Writing.io Jobs

Research Manager / Sr. Manager, Custom Insights & Strategy

Location: Remote

Overview

Collage Group is the only cultural intelligence engine that fuses consumer, industry, and brand data to provide brands with insights on the why behind consumers’ behaviors – so they can act quickly to fuel revenue and brand growth. Powered by 28B points of data and AI workflows, we’re helping 250+ of the world’s largest brands win with key growth audiences.

Collage Group is seeking a Research Manager / Senior Research Manager to join our Custom Insights market research team in our mission to uncover hidden truths and immersive human insights that propel brands forward.

In this role, you will lead full service custom market research engagements across both Quantitative and Qualitative methodologies. You will serve as a strategic thought partner to major brand clients, synthesizing insights from primary data and Collage’s proprietary cultural intelligence assets to deliver actionable, high-impact stories that drive confidence in decision-making.

Note: This is a consumer market research position and not a business analyst, UX research, or evaluation research role.

Responsibilities

• Full Service Research Execution: Manage and execute custom quantitative and qualitative research studies from kickoff to presentation, championing research quality at every stage of the process. This includes ownership of tasks such as questionnaire and discussion guide design, field management, survey and data QC, data analysis, and reporting (quant and qual data visualization in PowerPoint).
• Insight Generation & Storytelling: Synthesize both qualitative and quantitative data into strategic research reports with compelling narratives to inspire clients to action. Serve as the representative of the consumer voice, delivering culturally fluent insights on under-represented audiences and an empathetic perspective on consumer tensions.
• Project Management: Oversee project timelines and task execution to ensure projects are completed on time, meet research objectives, and drive client delight and loyalty.
• Team Collaboration: Partner closely with project team members and vendors to ensure alignment on task division, timelines, and project objectives. Help the team proactively problem-solve by flagging any capacity, quality, or timing issues early. Provide direction and mentorship to junior analysts on project teams (e.g., slide building best practices). Note: This role does not have direct reports/management responsibilities.
• Client Servicing: Serve as the client contact on key projects or support project lead on client management and communication. Lead project kick off calls and presentations, ensuring research objectives and core business questions are addressed with meaningful insights and brand-specific recommendations.

Experience

• 5-8+ years demonstrated experience in consumer market research (agency, supplier, or consultancy side preferred)
• Strong hands-on proficiency in both quantitative (survey design, statistical analysis, data visualization) and qualitative research techniques (discussion guide design, online bulletin boards, IDIs, focus groups, qualitative data analysis & reporting)
• Proven track record of creating visually compelling, action-oriented presentation decks that translate complex data into clear recommendations for non-researchers and senior executive audiences
• Experience directly working with clients (e.g., presenting research findings to key client stakeholders)
• Intellectually curious, with a passion for consumer culture, human behavior, and brainstorming insights and creative research design
• Highly organized, able to manage multiple tasks or projects concurrently, strong follow-through on deadlines, and a passion for high-quality research
• Equally comfortable in a collaborative, co-creative team environment and with self-directed responsibilities
• Advanced proficiency in Excel & PowerPoint
• Experience with survey platforms (e.g., Decipher, Qualtrics) and qualitative platforms (e.g., Recollective, Qualboard, Discuss.io), analytical/statistical tools a plus (e.g., SPSS, R)
• Spanish language proficiency a plus

Target Compensation:$75,000 - $100,000 + bonus

Invested in Your Well-Being:

We don’t just hire employees; we invest in people. Your health is our priority, which is why we offer premium medical, dental, and vision coverage for you and your loved ones. We believe in a life outside of work, supported by a generous PTO program that includes 12 company holidays, a floating holiday, and a dedicated day for you to give back to a cause you love. We’re also proud to support growing families with 8 weeks of fully paid parental leave for all parents, regardless of path to parenthood. Finally, we help you build for tomorrow with a competitive 401(k) match—because your future is just as important as your present.

Details regarding our Benefits can be found on our Careers Page.

All full time direct employees must be authorized to work in the United States, and must be US-based. Please note that while we are happy to hire remote unless otherwise stated in the job description, we are only able to employ people in the following US States: CA, CO, CT, FL, GA, IL, IN, KS, MD, MA, MN, MO, NJ, NY, NC, OH, PA, RI, TN, TX, VA and DC.

Important: If you have not interviewed directly with an authorized representative of Collage and you receive a job offer, do not respond or provide any personal information. This is likely a fraudulent attempt. Collage will never request sensitive information (such as your Social Security number, banking details, or passwords) before a formal interview process.

About Us: Collage is the only AI-enabled cultural intelligence engine that provides brands with curated, actionable, cultural insights to accelerate revenue growth. Only Collage fuses consumer, brand and industry insights from 26 billion primary data points to provide brands with unparalleled depth of insights on the why behind consumers’ behaviors. The combination of the always-on, easy to use, updated daily tech platform with the best-in-class cultural strategists provides a unique competitive advantage for the world’s leading brands. Collage enables marketers and insights professionals to futureproof their brands, identify and win with the fastest-growing segments, increase brand resonance and loyalty, and authentically activate across all the marketing levers. Founded in 2009, culture has always been at the company’s core. Collage is a National Minority Supplier Development Council (NMSDC) certified minority-owned small business.

Collage Group provides equal employment opportunities to all employees and applicants for employment and prohibits discrimination and harassment of any type without regard to race, color, religion, age, sex, national origin, disability status, genetics, protected veteran status, sexual orientation, gender identity or expression, or any other characteristic protected by federal, state, or local laws.

The Personalization team makes deciding what to play next easier and more enjoyable for every listener. From Blend to Discover Weekly, we’re behind some of Spotify’s most-loved features. We built...

Welcome to DechraDechra is an international veterinary pharmaceutical business focused on specialty products.

About the Company

The mining industry has steadily become worse at finding new ore deposits, requiring >10X more capital to make discoveries compared to 30 years ago. The easy-to-find, near-surface deposits have largely been found, and the industry has chronically under-invested in new exploration technology, relying on the manual techniques of yesteryear – even as demand accelerates for copper, lithium, and other metals to build electric vehicles, renewable energy, and data centers.

KoBold builds AI models for mineral exploration and deploys those models—alongside our novel sensors—to guide decisions on KoBold-owned-and-operated exploration programs. In the eight years since founding, KoBold has become by far both the largest independent mineral exploration company and the largest exploration technology developer. Our data scientists and software engineers, who come from leading technology companies, jointly lead exploration programs with our renowned exploration geologists.

KoBold has proven its first discovery with materially less capital than the industry average and found one of the best copper deposits ever discovered: the copper is far more concentrated than the global average of copper mines, and this asset alone is expected to generate meaningful revenue for decades. KoBold has a portfolio of more than 80 other projects, each of which has the potential for another high-quality discovery.

KoBold is privately held; investors include institutional asset managers T. Rowe Price and Canada Pension Plan Investments; technology venture capitalists Andreessen Horowitz, Breakthrough Energy Ventures, BOND Capital, Durable Capital, StepStone, and Standard Investments; and strategic natural resources companies BHP and Mitsubishi.

About the Role

KoBold explores across several continents and a widening set of commodities. This role is the geochemical capability that travels with those programs.

Reporting to the Principal Geochemist, you will work as a functioning member of exploration teams in North America, Europe, and Africa; planning and running sampling campaigns, on the ground when it matters, and turning results around quickly enough to change what the team does next.

The Principal Geochemist has responsibility for company-wide standards, independent data quality assurance, and global method and technology development. Your responsibility is in the practice, programs, field decisions, interpretation, and the day-to-day geochemical judgement that the exploration teams you work within depend on. You will work closely with data scientists and software engineers to develop new tools and drive an ever deepening analysis of results. You will work to company standards and audit against them in the field. You will be required to prioritize your time effectively across multiple projects, and multiply your reach by training field teams to sample well without you present.

Responsibilities

• Demonstrate in all activities a commitment to the highest standards in health and safety, environment, and community relations
• Act as the embedded geochemist for exploration programs across North America, Europe, and Africa, joining project teams for the phases where geochemical decisions are being made and providing insight to existing hypotheses and the generation of new hypotheses fast enough to influence exploration decision making
• Design geochemical sampling programs fit for the terrain, regolith, and mineral system being tested, including orientation surveys and the choice of sampling medium, horizon, size fraction, spacing, analytical suite, and digestion, across glaciated terrain, deeply weathered tropical profiles, and arid or transported cover
• Own the geochemical component of drilling programs: sampling intervals and sample support, preparation protocols, quality control insertion, and real-time review of incoming assays against the geological model
• Apply and audit KoBold’s geochemical standards and quality control procedures in the field, resolve routine control failures at program level, and escalate systemic or analytical problems to the Principal Geochemist, who retains independent data quality sign-off
• In close collaboration with the science team, interpret multi-element geochemical datasets across commodities and deposit types, including lithogeochemical classification, alteration and trace element vectoring, fertility screening, and appropriate treatment of compositional data
• Partner with geoscientists, data scientists and software engineers to improve existing tools, and develop new tools
• Assess legacy and third-party geochemical datasets during property evaluation and acquisition due diligence, judging quality, comparability, and fitness for purpose, and identifying where re-analysis or re-sampling is required before the data can carry weight
• Work alongside geologists, geophysicists, and data scientists to identify, rank, and prioritise targets, making geochemistry a primary input to targeting
• Coach and train field geologists and technicians in leading-practice sampling, and contribute to the standard operating procedures and field documentation that let sampling quality hold when no geochemist is on site
• Feed field experience back into method and technology development led by the Principal Geochemist, including practical trials of portable XRF, in-field sample preparation, and emerging analytical techniques
• Plan and manage a demanding international travel and deployment schedule, in coordination with the Principal Geochemist and project leadership

Qualifications

A great Senior Geochemist candidate will have:

• Strong technical knowledge of more than one ore deposit type and comfort moving between mineral systems. Experience across two or more of sediment-hosted stratiform copper, porphyry copper, IOCG, magmatic nickel-copper-PGE sulfide, or LCT pegmatite systems will be considered an advantage
• Demonstrated hands-on experience designing and executing surface geochemical programs, including orientation work, in more than one regolith or cover setting
• Experience running the geochemical side of drilling programs, from sampling protocol through quality control to interpretation of results as they arrive
• Practical command of quality control for sampling programs, including reference material and duplicate design, control charting, and precision and bias assessment, together with working knowledge of the Theory of Sampling
• Excellence in generative analysis, from new concepts through to the integration, interpretation, and modeling of large multi-element datasets
• Baseline level of proficiency in standard desktop software packages (e.g., ArcGIS or QGIS) and in at least one dedicated geochemical data analysis package (e.g., ioGAS)
• The ability to work effectively as a member of teams they do not manage, across cultures, time zones, and technical disciplines, and to earn influence through the quality of their judgement rather than through reporting lines
• Previous experience in coaching and training geoscientists of varying experience, educational backgrounds and cultures in leading-practice mineral exploration
• Willingness and ability to travel internationally for extended periods, and eligibility to work or obtain the necessary permissions across the regions covered by this role
• Demonstrated commitment to high standards of health and safety in the workplace
• A minimum of 8 years’ experience in the mineral exploration industry
• Intellectual curiosity and eagerness to learn about all aspects of mineral exploration, particularly in the geology domain. Enjoys constantly learning such that you are driving insights through using our tools in exploration and willing to work directly with geologists in the field.
• Ability to explain technical problems to and collaborate on solutions with domain experts who are not Geoscientists. A strong communicator who enjoys working with colleagues across the company.
• Excitement about joining a fast-growing early-stage company, comfort with a dynamic work environment, and eagerness to take on an evolving range of responsibilities.
• Keen not just to build cool technology, but to figure out what technical product to build to best achieve the business objectives of the company.
• A valid passport and willingness to travel to exploration sites around the world, especially during active field seasons.

It is helpful but not required to have:

• Working knowledge of one programming or scripting language (e.g., Python, R) applied to geochemical data
• Knowledge of geostatistics, compositional data analysis, or other statistical methods relevant for analysis of geoscientific data
• Experience with portable XRF and in-field sample preparation as auditable, fit-for-purpose analytical capability
• Experience in environmental geochemistry, including baseline surveys and waste and tailings characterisation
• International experience, particularly in the Central African Copperbelt, or in the pegmatite provinces of central Africa and Namibia
• Professional registration or chartership (e.g., CGeol, FAusIMM, EurGeol, SACNASP)
• Working proficiency in French

Note: Candidates whose background is in exploration geology with a strong geochemical specialisation, and who want to work as a geochemist across a global portfolio rather than a single project, are encouraged to apply.

KoBold Metals is an equal opportunity workplace and an affirmative action employer. We are committed to equal employment opportunities for people of any race, color, ancestry, religion, sex, gender identity, sexual orientation, marital status, national origin, age, citizenship, disability, or veteran status.

This position is Full-time Exempt.

The US base salary range for this full-time exempt position is $140,000-$185,000, plus equity and benefits including 401k matching.

Location: KoBold is a remote-first workplace. All candidates must be legally authorized to work in one of the following countries: United States, Canada, Australia, South Africa, United Kingdom, Zambia, Democratic Republic of the Congo, Namibia, Finland, Sweden, or Botswana.

Please submit your application to: https://koboldmetals.com/careers/

What to Expect

Joining KoBold means getting the opportunity for hands-on exposure to our exploration projects around the world. All employees are expected to travel to project sites, with a minimum of one week per year. Field-facing and technical roles spend significantly more time in the field.

Compensation ranges are listed in USD. If you are based outside the US, the range may not translate directly: US pay reflects a different market with different competitive dynamics, so we benchmark roles against the local market rather than converting the listed figure.

About the Company

The mining industry has steadily become worse at finding new ore deposits, requiring >10X more capital to make discoveries compared to 30 years ago. The easy-to-find, near-surface deposits have largely been found, and the industry has chronically under-invested in new exploration technology, relying on the manual techniques of yesteryear – even as demand accelerates for copper, lithium, and other metals to build electric vehicles, renewable energy, and data centers.

KoBold builds AI models for mineral exploration and deploys those models—alongside our novel sensors—to guide decisions on KoBold-owned-and-operated exploration programs. In the eight years since founding, KoBold has become by far both the largest independent mineral exploration company and the largest exploration technology developer. Our data scientists and software engineers, who come from leading technology companies, jointly lead exploration programs with our renowned exploration geologists.

KoBold has proven its first discovery with materially less capital than the industry average and found one of the best copper deposits ever discovered: the copper is far more concentrated than the global average of copper mines, and this asset alone is expected to generate meaningful revenue for decades. KoBold has a portfolio of more than 80 other projects, each of which has the potential for another high-quality discovery.

KoBold is privately held; investors include institutional asset managers T. Rowe Price and Canada Pension Plan Investments; technology venture capitalists Andreessen Horowitz, Breakthrough Energy Ventures, BOND Capital, Durable Capital, StepStone, and Standard Investments; and strategic natural resources companies BHP and Mitsubishi.

About the Role

KoBold explores across several continents and a widening set of commodities. This role is the geochemical capability that travels with those programs.

Reporting to the Principal Geochemist, you will work as a functioning member of exploration teams in North America, Europe, and Africa; planning and running sampling campaigns, on the ground when it matters, and turning results around quickly enough to change what the team does next.

The Principal Geochemist has responsibility for company-wide standards, independent data quality assurance, and global method and technology development. Your responsibility is in the practice, programs, field decisions, interpretation, and the day-to-day geochemical judgement that the exploration teams you work within depend on. You will work closely with data scientists and software engineers to develop new tools and drive an ever deepening analysis of results. You will work to company standards and audit against them in the field. You will be required to prioritize your time effectively across multiple projects, and multiply your reach by training field teams to sample well without you present.

Responsibilities

• Demonstrate in all activities a commitment to the highest standards in health and safety, environment, and community relations
• Act as the embedded geochemist for exploration programs across North America, Europe, and Africa, joining project teams for the phases where geochemical decisions are being made and providing insight to existing hypotheses and the generation of new hypotheses fast enough to influence exploration decision making
• Design geochemical sampling programs fit for the terrain, regolith, and mineral system being tested, including orientation surveys and the choice of sampling medium, horizon, size fraction, spacing, analytical suite, and digestion, across glaciated terrain, deeply weathered tropical profiles, and arid or transported cover
• Own the geochemical component of drilling programs: sampling intervals and sample support, preparation protocols, quality control insertion, and real-time review of incoming assays against the geological model
• Apply and audit KoBold’s geochemical standards and quality control procedures in the field, resolve routine control failures at program level, and escalate systemic or analytical problems to the Principal Geochemist, who retains independent data quality sign-off
• In close collaboration with the science team, interpret multi-element geochemical datasets across commodities and deposit types, including lithogeochemical classification, alteration and trace element vectoring, fertility screening, and appropriate treatment of compositional data
• Partner with geoscientists, data scientists and software engineers to improve existing tools, and develop new tools
• Assess legacy and third-party geochemical datasets during property evaluation and acquisition due diligence, judging quality, comparability, and fitness for purpose, and identifying where re-analysis or re-sampling is required before the data can carry weight
• Work alongside geologists, geophysicists, and data scientists to identify, rank, and prioritise targets, making geochemistry a primary input to targeting
• Coach and train field geologists and technicians in leading-practice sampling, and contribute to the standard operating procedures and field documentation that let sampling quality hold when no geochemist is on site
• Feed field experience back into method and technology development led by the Principal Geochemist, including practical trials of portable XRF, in-field sample preparation, and emerging analytical techniques
• Plan and manage a demanding international travel and deployment schedule, in coordination with the Principal Geochemist and project leadership

Qualifications

A great Senior Geochemist candidate will have:

• Strong technical knowledge of more than one ore deposit type and comfort moving between mineral systems. Experience across two or more of sediment-hosted stratiform copper, porphyry copper, IOCG, magmatic nickel-copper-PGE sulfide, or LCT pegmatite systems will be considered an advantage
• Demonstrated hands-on experience designing and executing surface geochemical programs, including orientation work, in more than one regolith or cover setting
• Experience running the geochemical side of drilling programs, from sampling protocol through quality control to interpretation of results as they arrive
• Practical command of quality control for sampling programs, including reference material and duplicate design, control charting, and precision and bias assessment, together with working knowledge of the Theory of Sampling
• Excellence in generative analysis, from new concepts through to the integration, interpretation, and modeling of large multi-element datasets
• Baseline level of proficiency in standard desktop software packages (e.g., ArcGIS or QGIS) and in at least one dedicated geochemical data analysis package (e.g., ioGAS)
• The ability to work effectively as a member of teams they do not manage, across cultures, time zones, and technical disciplines, and to earn influence through the quality of their judgement rather than through reporting lines
• Previous experience in coaching and training geoscientists of varying experience, educational backgrounds and cultures in leading-practice mineral exploration
• Willingness and ability to travel internationally for extended periods, and eligibility to work or obtain the necessary permissions across the regions covered by this role
• Demonstrated commitment to high standards of health and safety in the workplace
• A minimum of 8 years’ experience in the mineral exploration industry
• Intellectual curiosity and eagerness to learn about all aspects of mineral exploration, particularly in the geology domain. Enjoys constantly learning such that you are driving insights through using our tools in exploration and willing to work directly with geologists in the field.
• Ability to explain technical problems to and collaborate on solutions with domain experts who are not Geoscientists. A strong communicator who enjoys working with colleagues across the company.
• Excitement about joining a fast-growing early-stage company, comfort with a dynamic work environment, and eagerness to take on an evolving range of responsibilities.
• Keen not just to build cool technology, but to figure out what technical product to build to best achieve the business objectives of the company.
• A valid passport and willingness to travel to exploration sites around the world, especially during active field seasons.

It is helpful but not required to have:

• Working knowledge of one programming or scripting language (e.g., Python, R) applied to geochemical data
• Knowledge of geostatistics, compositional data analysis, or other statistical methods relevant for analysis of geoscientific data
• Experience with portable XRF and in-field sample preparation as auditable, fit-for-purpose analytical capability
• Experience in environmental geochemistry, including baseline surveys and waste and tailings characterisation
• International experience, particularly in the Central African Copperbelt, or in the pegmatite provinces of central Africa and Namibia
• Professional registration or chartership (e.g., CGeol, FAusIMM, EurGeol, SACNASP)
• Working proficiency in French

Note: Candidates whose background is in exploration geology with a strong geochemical specialisation, and who want to work as a geochemist across a global portfolio rather than a single project, are encouraged to apply.

KoBold Metals is an equal opportunity workplace and an affirmative action employer. We are committed to equal employment opportunities for people of any race, color, ancestry, religion, sex, gender identity, sexual orientation, marital status, national origin, age, citizenship, disability, or veteran status.

This position is Full-time Exempt.

The US base salary range for this full-time exempt position is $140,000-$185,000, plus equity and benefits including 401k matching.

Location: KoBold is a remote-first workplace. All candidates must be legally authorized to work in one of the following countries: United States, Canada, Australia, South Africa, United Kingdom, Zambia, Democratic Republic of the Congo, Namibia, Finland, Sweden, or Botswana.

Please submit your application to: https://koboldmetals.com/careers/

What to Expect

Joining KoBold means getting the opportunity for hands-on exposure to our exploration projects around the world. All employees are expected to travel to project sites, with a minimum of one week per year. Field-facing and technical roles spend significantly more time in the field.

Compensation ranges are listed in USD. If you are based outside the US, the range may not translate directly: US pay reflects a different market with different competitive dynamics, so we benchmark roles against the local market rather than converting the listed figure.

This is an exciting opportunity to advance Natera’s regulated product portfolio of cutting-edge molecular diagnostic tests. As a Staff Biostatistician, you will leverage your strong applied statistics, IVD, and Clinical Research expertise to drive the strategy, design, analysis, and interpretation of clinical validation studies for kidney disease and other high-priority germline projects.

PRIMARY RESPONSIBILITIES:

• Strategic Study Design and Analysis: Independently lead the design, analysis, and interpretation of clinical validation studies to support in vitro diagnostic approvals (e.g., CDx, 510(k), PMA, etc.)
• Regulatory Compliance: Ensure statistical methodologies and analyses meet the standards set by regulatory agencies such as FDA. Drive statistical interactions with regulatory authorities.
• Documentation and Reporting: Prepare comprehensive statistical analysis protocols (SAPs), reports, and presentations for cross-functional and external collaborators and stakeholders.
• Cross-Functional Leadership: Serve as the statistical lead on high-priority project teams. Partner with Clinical Science, Data Management, and Product teams to integrate statistical expertise into project planning and execution. Identify risks, communicate concerns, and champion solutions to interdisciplinary project teams
• Clinical Research and Beyond: Drive design and execution of high-impact research clinical trials, studies, and publications that push the development of our world-class product portfolio
• Stay current with advancements in Biostatistics and clinical validation methodology, implementing innovative approaches to support diagnostic product development and regulatory submissions
• Mentor and provide technical supervision to junior team members

QUALIFICATIONS:

• Education: Ph.D. or Master’s degree in Biostatistics, Statistics, or a related field
• Experience:
  • At least 8 years of relevant industry experience, preferably within molecular diagnostics, biotechnology, or related fields.
  • Demonstrated experience in regulated environments, including previous CDx or IVD submission experience
  • Experience with next-generation sequencing (NGS), genetic, and/or germline-related data.
  • Minimum of 4 years practical experience with R statistical analysis software

KNOWLEDGE, SKILLS, AND ABILITIES:

• Expert working knowledge of statistical methodology in diagnostic medicine, including an understanding of FDA guidelines
• Previous experience interacting with regulatory bodies such as the FDA.
• Demonstrated ability and enthusiasm for working on cross-functional teams with members of diverse technical backgrounds
• Strong track record of producing high quality written documentation for multiple audiences
• Prior experience with germline assay development preferred
• Prior experience with kidney disease desired
• Experience in design and analyses of clinical validation studies
• Strong communication skills, with the ability to explain complex statistical concepts to non-statistical audiences.
• Attention to detail and a commitment to delivering high-quality work under tight deadlines.

The pay range is listed and actual compensation packages are based on a wide array of factors unique to each candidate, including but not limited to skill set, years & depth of experience, certifications and specific office location. This may differ in other locations due to cost of labor considerations.

Remote USA

$164,000—$205,000 USD

OUR OPPORTUNITY

Natera™ is a global leader in cell-free DNA (cfDNA) testing, dedicated to oncology, women’s health, and organ health. Our aim is to make personalized genetic testing and diagnostics part of the standard of care to protect health and enable earlier and more targeted interventions that lead to longer, healthier lives.

The Natera team consists of highly dedicated statisticians, geneticists, doctors, laboratory scientists, business professionals, software engineers and many other professionals from world-class institutions, who care deeply for our work and each other. When you join Natera, you’ll work hard and grow quickly. Working alongside the elite of the industry, you’ll be stretched and challenged, and take pride in being part of a company that is changing the landscape of genetic disease management.

WHAT WE OFFER

Competitive Benefits - Employee benefits include comprehensive medical, dental, vision, life and disability plans for eligible employees and their dependents. Additionally, Natera employees and their immediate families receive free testing in addition to fertility care benefits. Other benefits include pregnancy and baby bonding leave, 401k benefits, commuter benefits and much more. We also offer a generous employee referral program!

For more information, visit www.natera.com.

Natera is proud to be an Equal Opportunity Employer. We are committed to ensuring a diverse and inclusive workplace environment, and welcome people of different backgrounds, experiences, abilities and perspectives. Inclusive collaboration benefits our employees, our community and our patients, and is critical to our mission of changing the management of disease worldwide.

All qualified applicants are encouraged to apply, and will be considered without regard to race, color, religion, gender, gender identity or expression, sexual orientation, national origin, genetics, age, veteran status, disability or any other legally protected status. We also consider qualified applicants regardless of criminal histories, consistent with applicable laws.

If you are based in California, we encourage you to read this important information for California residents.

Link: https://www.natera.com/notice-of-data-collection-california-residents/

Please be advised that Natera will reach out to candidates with a @natera.com email domain ONLY. Email communications from all other domain names are not from Natera or its employees and are fraudulent. Natera does not request interviews via text messages and does not ask for personal information until a candidate has engaged with the company and has spoken to a recruiter and the hiring team. Natera takes cyber crimes seriously, and will collaborate with law enforcement authorities to prosecute any related cyber crimes.

For more information:

- BBB announcement on job scams

- FBI Cyber Crime resource page

The Personalization team makes deciding what to play next easier and more enjoyable for every listener. From Blend to Discover Weekly, we’re behind some of Spotify’s most-loved features. We built...

Description

Join Tether and Shape the Future of Digital Finance

At Tether, we’re not just building products, we’re pioneering a global financial revolution. Our cutting-edge solutions empower businesses—from exchanges and wallets to payment processors and ATMs—to seamlessly integrate reserve-backed tokens across blockchains. By harnessing the power of blockchain technology, Tether enables you to store, send, and receive digital tokens instantly, securely, and globally, all at a fraction of the cost. Transparency is the bedrock of everything we do, ensuring trust in every transaction.

Innovate with Tether

Tether Finance: Our innovative product suite features the world’s most trusted stablecoin, USDT, relied upon by hundreds of millions worldwide, alongside pioneering digital asset tokenization services.

But that’s just the beginning:

Tether Power: Driving sustainable growth, our energy solutions optimize excess power for Bitcoin mining using eco-friendly practices in state-of-the-art, geo-diverse facilities.

Tether Data: Fueling breakthroughs in AI and peer-to-peer technology, we reduce infrastructure costs and enhance global communications with cutting-edge solutions like KEET, our flagship app that redefines secure and private data sharing.

Tether Education: Democratizing access to top-tier digital learning, we empower individuals to thrive in the digital and gig economies, driving global growth and opportunity.

Tether Evolution: At the intersection of technology and human potential, we are pushing the boundaries of what is possible, crafting a future where innovation and human capabilities merge in powerful, unprecedented ways.

Why Join Us?

Our team is a global talent powerhouse, working remotely from every corner of the world. If you’re passionate about making a mark in the fintech space, this is your opportunity to collaborate with some of the brightest minds, pushing boundaries and setting new standards. We’ve grown fast, stayed lean, and secured our place as a leader in the industry.

If you have excellent English communication skills and are ready to contribute to the most innovative platform on the planet, Tether is the place for you.

Are you ready to be part of the future?

About the job

As a member of our AI research team, you will drive innovation in model compression and efficient deployment for advanced multimodal AI systems, including large language models (LLMs) and vision-language models (VLMs). Your work will focus on reducing model footprint and computational cost while preserving accuracy, enabling high-performance AI to run efficiently across resource-constrained edge devices. You will apply and advance compression techniques such as quantization, knowledge distillation, and pruning to streamline complex multimodal architectures that integrate text, images, and audio.

We expect you to have deep expertise in model compression methods and a strong background in multimodal model architectures. You will adopt a hands-on, research-driven approach to develop, test, and implement novel compression strategies that balance model size, latency, throughput, and accuracy. Your responsibilities include building robust compression pipelines, establishing performance and fidelity metrics, and addressing bottlenecks in production inference. The ultimate goal is to deliver scalable, low-memory, low-latency AI systems on edge devices (i.e., smartphones) that maintain high fidelity and tangible real-world value.

Responsibilities

Break down the key responsibilities in bullet points. It’s helpful to make them actionable and measurable. This could also be grouped into categories for more complex roles.

• Apply low-bit quantization to reduce model size and inference latency for generative AI models (LLMs, VLMs, multimodal) while maintaining accuracy and output quality.

• Leverage knowledge distillation to transfer capabilities from larger teacher models to smaller student models, enabling efficient multimodal reasoning across text, image, and audio inputs.

• Implement pruning techniques to remove redundant parameters and attention heads, reducing computational overhead without sacrificing task performance.

• Analyze trade-offs between model efficiency (size, latency, memory) and accuracy across quantization, distillation, and pruning methods; propose improvements based on empirical findings.

• Research and apply mixed-precision quantization and other advanced compression strategies (e.g., adaptive pruning schedules, distillation with intermediate feature matching) to optimize the accuracy–performance balance.

• Stay current with the latest research in model compression, including emerging techniques for multimodal and generative architectures.

• Document methodologies, experiments, and results clearly to support reproducibility, internal collaboration, and stakeholder communication.

• Author technical papers and publish findings in top-tier conferences (e.g., NeurIPS, ICML, ICLR, CVPR, ACL, AAAI) to advance the field of model compression for multimodal AI.

Requirements

List the qualifications, certifications, and experience needed for the role. You can split this into mandatory and preferred sections if needed.

• A degree in Computer Science  or related field. Ideally PhD in NLP, Machine Learning, or a related field, complemented by a solid track record in AI R&D (with good publications in A* conferences).

• Experience with PyTorch deep learning frameworks or equivalent frameworks

• Hands-on experience with model quantization including both Quantization-Aware Training (QAT) and Post-Training Quantization (PTQ).

• Research and hands-on experience with knowledge distillation for compressing large models into smaller, efficient ones.

• Research and hands-on experience with model pruning for compressing large models into smaller, efficient ones.

• Solid understanding of neural network architectures and training processes – Including transformers (e.g., LLMs, VLMs), backpropagation, optimization, and fine-tuning techniques.

• Familiarity with C++ is a plus (especially for implementing low-level quantization kernels or inference optimizations).

Important information for candidates

Recruitment scams have become increasingly common. To protect yourself, please keep the following in mind when applying for roles:

• Apply only through our official channels. We do not use third-party platforms or agencies for recruitment unless clearly stated. All open roles are listed on our official careers page: https://tether.recruitee.com/

• Verify the recruiter’s identity. All our recruiters have verified LinkedIn profiles. If you’re unsure, you can confirm their identity by checking their profile or contacting us through our website.

• Be cautious of unusual communication methods. We do not conduct interviews over WhatsApp, Telegram, or SMS. All communication is done through official company emails and platforms.

• Double-check email addresses. All communication from us will come from emails ending in @ tether.to or @ tether.io

• We will never request payment or financial details. If someone asks for personal financial information or payment at any point during the hiring process, it is a scam. Please report it immediately.

When in doubt, feel free to reach out through our official website.

Telix Pharmaceuticals is a dynamic, fast-growing radiopharmaceutical company with offices in Australia (HQ), Belgium, Japan, Switzerland, and the United States. Our mission is to deliver on the promise of precision medicine through targeted radiation. At our core, we exist to create products that seek to improve the quality of life for people living with cancer and rare diseases. You will be supporting the international roll-out of our approved prostate cancer imaging agent and helping us to advance our portfolio of late clinical-stage products that address significant unmet need in oncology and rare diseases.

See Yourself at Telix

This role provides technical and analytical expertise for early phase biologics process development and manufacturing activities for multiple technology areas (e.g. analytical, purification, formulation, and bioconjugation) ensuring manufacturing processes are sufficiently developed to meet program objectives, safety standards, regulatory expectations, and quality compliance. This role is responsible for managing projects in multiple technology areas (e.g. analytical, purification, formulation, or bioconjugation) at internal sites and external CMC partners, as well as contributing to internal cross-functional collaborations to develop and execute clinical manufacturing.

The Principal Scientist, Biologics acts as the Biologics lead for a specific therapeutic program and should be comfortable communicating with cross-functional teams in formal settings and third-party meetings, be knowledgeable in several technical areas as well as their focus technology area (e.g. analytical, purification, formulation, or bioconjugation), and manage their own workload to complete assignments autonomously with some feedback from senior-level colleagues, matrix team members,

and/or their manager for more complicated topics.

Key Accountabilities:

• Leads development projects for multiple technology areas (e.g. analytical, purification, formulation, or bioconjugation) ensuring the appropriate process knowledge, technology transfer, analytical method transfer, design space and/or control strategy definition, optimization, and/or qualification to produces clinical drug substance intermediates.
• Supports review and approval of manufacturing operation documents such as SOPs, batch records, protocols, analytical procedures, etc., and development reports for multiple technology areas (e.g. analytical, purification, formulation, or bioconjugation) to produce clinical drug substance intermediates.
• Ensures appropriate compliance of multiple technology areas (e.g. analytical, purification, formulation, or bioconjugation) through document creation/review/approval, quality investigations, corrective actions, and change control within a Quality system.
• Authors and/or reviews source documents and CMC content related to several technology areas (e.g. analytical, purification, formulation, or bioconjugation) to support IND, NDA, BLA and similar regulatory agency applications, especially in areas requiring more sophisticated acumen such as comparability and risk-based approach
• Leads collaborations with internal sites and external CMC partners to develop selection criteria, assess capabilities, coordinate activities, plan process technology /analytical method activities, troubleshoot and analyze operational performance, conduct risk assessments, and manage timely deliverables.

Mentors Senior Scientists and other level Scientists on both technical topics as well as Telix values.

Education and Experience:

• Science or engineering degree with minimum relevant experience as applicable: BS with 11 years of experience, MS with 9 years of experience, PhD with 8 years of experience.
• Minimum 5 years of experience in GMP, quality management system, and CMC compliance and
• Demonstrated understanding and application of project management principles and
• Ability to work in cross-functional, virtual
• Travel expectation: 10 – 25% for key in-person meetings, partner site visits, and SME/technical person-in-plant

Key Capabilities:

• Commitment to excellence: Take pride in your work and consistently strive for excellence in everything you do
• Results-oriented: Driven to achieve goals and objectives, with a strong focus on delivering measurable results
• Creativity and innovation: Possess a willingness to think outside the box and come up with unique and creative solutions to challenges
• Strong communication skills: Able to communicate effectively with colleagues and stakeholders at all levels, using clear and concise language
• Collaboration: Work effectively as part of a team, actively sharing knowledge and expertise to achieve common goals
• Resilience: Demonstrate the ability to bounce back from setbacks and persevere in the face of challenges
• Inclusive mindset: Demonstrate an understanding and appreciation for diversity, and actively work to create an inclusive environment where everyone feels valued and respected
• Adaptability: Comfortable working in a dynamic environment, able to adjust to changing priorities, and willing to take on new challenges
• Ethical behavior: Act with integrity and demonstrate a commitment to ethical behavior in all interactions with colleagues and stakeholders
• Continuous learning: Show a commitment to ongoing learning and professional development, continually seeking out opportunities to expand your knowledge and skills

At Telix, we believe everyone counts, we strive to be extraordinary, and we pursue our goals with determination and integrity. You will be part of an engaged and supportive group of colleagues who all have a shared purpose: to help people with cancer and rare diseases live longer, better quality lives. Being a Telix employee means being a part of a unique, global, interdisciplinary team working to deliver what’s next in precision medicine. We have hybrid and remote employees located all around the world. We offer competitive salaries, annual performance-based bonuses, an equity-based incentive program, generous vacation, paid wellness days and support for learning and development.

Telix Pharmaceuticals is an equal opportunity employer. We evaluate qualified applicants without regard to race, color, religion, sex, sexual orientation, gender identity, national origin, disability, veteran status, and other legally protected characteristics.

VIEW OUR PRIVACY POLICY HERE

Telix Pharmaceuticals is a dynamic, fast-growing radiopharmaceutical company with offices in Australia (HQ), Belgium, Japan, Switzerland, and the United States. Our mission is to deliver on the promise of precision medicine through targeted radiation. At our core, we exist to create products that seek to improve the quality of life for people living with cancer and rare diseases. You will be supporting the international roll-out of our approved prostate cancer imaging agent and helping us to advance our portfolio of late clinical-stage products that address significant unmet need in oncology and rare diseases.

See Yourself at Telix

This role provides technical and analytical expertise for early phase biologics process development and manufacturing activities for multiple technology areas (e.g. analytical, purification, formulation, and bioconjugation) ensuring manufacturing processes are sufficiently developed to meet program objectives, safety standards, regulatory expectations, and quality compliance. This role is responsible for managing projects in multiple technology areas (e.g. analytical, purification, formulation, or bioconjugation) at internal sites and external CMC partners, as well as contributing to internal cross-functional collaborations to develop and execute clinical manufacturing.

The Principal Scientist, Biologics acts as the Biologics lead for a specific therapeutic program and should be comfortable communicating with cross-functional teams in formal settings and third-party meetings, be knowledgeable in several technical areas as well as their focus technology area (e.g. analytical, purification, formulation, or bioconjugation), and manage their own workload to complete assignments autonomously with some feedback from senior-level colleagues, matrix team members,

and/or their manager for more complicated topics.

Key Accountabilities:

• Leads development projects for multiple technology areas (e.g. analytical, purification, formulation, or bioconjugation) ensuring the appropriate process knowledge, technology transfer, analytical method transfer, design space and/or control strategy definition, optimization, and/or qualification to produces clinical drug substance intermediates.
• Supports review and approval of manufacturing operation documents such as SOPs, batch records, protocols, analytical procedures, etc., and development reports for multiple technology areas (e.g. analytical, purification, formulation, or bioconjugation) to produce clinical drug substance intermediates.
• Ensures appropriate compliance of multiple technology areas (e.g. analytical, purification, formulation, or bioconjugation) through document creation/review/approval, quality investigations, corrective actions, and change control within a Quality system.
• Authors and/or reviews source documents and CMC content related to several technology areas (e.g. analytical, purification, formulation, or bioconjugation) to support IND, NDA, BLA and similar regulatory agency applications, especially in areas requiring more sophisticated acumen such as comparability and risk-based approach
• Leads collaborations with internal sites and external CMC partners to develop selection criteria, assess capabilities, coordinate activities, plan process technology /analytical method activities, troubleshoot and analyze operational performance, conduct risk assessments, and manage timely deliverables.

Mentors Senior Scientists and other level Scientists on both technical topics as well as Telix values.

Education and Experience:

• Science or engineering degree with minimum relevant experience as applicable: BS with 11 years of experience, MS with 9 years of experience, PhD with 8 years of experience.
• Minimum 5 years of experience in GMP, quality management system, and CMC compliance and
• Demonstrated understanding and application of project management principles and
• Ability to work in cross-functional, virtual
• Travel expectation: 10 – 25% for key in-person meetings, partner site visits, and SME/technical person-in-plant

Key Capabilities:

• Commitment to excellence: Take pride in your work and consistently strive for excellence in everything you do
• Results-oriented: Driven to achieve goals and objectives, with a strong focus on delivering measurable results
• Creativity and innovation: Possess a willingness to think outside the box and come up with unique and creative solutions to challenges
• Strong communication skills: Able to communicate effectively with colleagues and stakeholders at all levels, using clear and concise language
• Collaboration: Work effectively as part of a team, actively sharing knowledge and expertise to achieve common goals
• Resilience: Demonstrate the ability to bounce back from setbacks and persevere in the face of challenges
• Inclusive mindset: Demonstrate an understanding and appreciation for diversity, and actively work to create an inclusive environment where everyone feels valued and respected
• Adaptability: Comfortable working in a dynamic environment, able to adjust to changing priorities, and willing to take on new challenges
• Ethical behavior: Act with integrity and demonstrate a commitment to ethical behavior in all interactions with colleagues and stakeholders
• Continuous learning: Show a commitment to ongoing learning and professional development, continually seeking out opportunities to expand your knowledge and skills

At Telix, we believe everyone counts, we strive to be extraordinary, and we pursue our goals with determination and integrity. You will be part of an engaged and supportive group of colleagues who all have a shared purpose: to help people with cancer and rare diseases live longer, better quality lives. Being a Telix employee means being a part of a unique, global, interdisciplinary team working to deliver what’s next in precision medicine. We have hybrid and remote employees located all around the world. We offer competitive salaries, annual performance-based bonuses, an equity-based incentive program, generous vacation, paid wellness days and support for learning and development.

Telix Pharmaceuticals is an equal opportunity employer. We evaluate qualified applicants without regard to race, color, religion, sex, sexual orientation, gender identity, national origin, disability, veteran status, and other legally protected characteristics.

VIEW OUR PRIVACY POLICY HERE

Thumbtack helps millions of people confidently care for their homes. Thumbtack is the one app you need to take care of and improve your home — from personalized guidance to...

Thumbtack helps millions of people confidently care for their homes. Thumbtack is the one app you need to take care of and improve your home — from personalized guidance to...

Company Overview

Deepgram is the leading platform underpinning the emerging trillion-dollar Voice AI economy, providing real-time APIs for speech-to-text (STT), text-to-speech (TTS), and building production-grade voice agents at scale. More than 200,000 developers and 1,300+ organizations build voice offerings that are ‘Powered by Deepgram’, including Twilio, Cloudflare, Sierra, Decagon, Vapi, Daily, Cresta, Granola, and Jack in the Box. Deepgram’s voice-native foundation models are accessed through cloud APIs or as self-hosted and on-premises software, with unmatched accuracy, low latency, and cost efficiency. Backed by a recent Series C led by leading global investors and strategic partners, Deepgram has processed over 50,000 years of audio and transcribed more than 1 trillion words. There is no organization in the world that understands voice better than Deepgram.

Company Operating Rhythm

At Deepgram, we expect an AI-first mindset—AI use and comfort aren’t optional, they’re core to how we operate, innovate, and measure performance.

Every team member who works at Deepgram is expected to actively use and experiment with advanced AI tools, and even build your own into your everyday work. We measure how effectively AI is applied to deliver results, and consistent, creative use of the latest AI capabilities is key to success here. Candidates should be comfortable adopting new models and modes quickly, integrating AI into their workflows, and continuously pushing the boundaries of what these technologies can do.

Additionally, we move at the pace of AI. Change is rapid, and you can expect your day-to-day work to evolve just as quickly. This may not be the right role if you’re not excited to experiment, adapt, think on your feet, and learn constantly, or if you’re seeking something highly prescriptive with a traditional 9-to-5.

The Opportunity

Voice is the most natural modality for human interaction with machines. However, current sequence modeling paradigms based on jointly scaling model and data cannot deliver voice AI capable of universal human interaction. The challenges are rooted in fundamental data problems posed by audio: real-world audio data is scarce and enormously diverse, spanning a vast space of voices, speaking styles, and acoustic conditions. Even if billions of hours of audio were accessible, its inherent high dimensionality creates computational and storage costs that make training and deployment prohibitively expensive at world scale. We believe that entirely new paradigms for audio AI are needed to overcome these challenges and make voice interaction accessible to everyone.

The Role

As a Member of the Research Staff, you will pioneer the development of Latent Space Models (LSMs), a new approach that aims to solve the fundamental data, scale, and cost challenges associated with building robust, contextualized voice AI. Your research will focus on solving one or more of the following problems:

• Build next-generation neural audio codecs that achieve extreme, low bit-rate compression and high fidelity reconstruction across a world-scale corpus of general audio.

• Pioneer steerable generative models that can synthesize the full diversity of human speech from the codec latent representation, from casual conversation to highly emotional expression to complex multi-speaker scenarios with environmental noise and overlapping speech.

• Develop embedding systems that cleanly factorize the codec latent space into interpretable dimensions of speaker, content, style, environment, and channel effects – enabling precise control over each aspect and the ability to massively amplify an existing seed dataset through “latent recombination”.

• Leverage latent recombination to generate synthetic audio data at previously impossible scales, unlocking joint model and data scaling paradigms for audio.  Endeavor to train multimodal speech-to-speech systems that can 1) understand any human irrespective of their demographics, state, or environment and 2) produce empathic, human-like responses that achieve conversational or task-oriented objectives.

• Design model architectures, training schemes, and inference algorithms that are adapted for hardware at the bare metal enabling cost efficient training on billion-hour datasets and powering real-time inference for hundreds of millions of concurrent conversations.

The Challenge

We are seeking researchers who:

• See “unsolved” problems as opportunities to pioneer entirely new approaches

• Can identify the one critical experiment that will validate or kill an idea in days, not months

• Have the vision to scale successful proofs-of-concept 100x

• Are obsessed with using AI to automate and amplify your own impact

If you find yourself energized rather than daunted by these expectations—if you’re already thinking about five ideas to try while reading this—you might be the researcher we need. This role demands obsession with the problems, creativity in approach, and relentless drive toward elegant, scalable solutions. The technical challenges are immense, but the potential impact is transformative.

It’s Important to Us That You Have

• Strong mathematical foundation in statistical learning theory, particularly in areas relevant to self-supervised and multimodal learning

• Deep expertise in foundation model architectures, with an understanding of how to scale training across multiple modalities

• Proven ability to bridge theory and practice—someone who can both derive novel mathematical formulations and implement them efficiently

• Demonstrated ability to build data pipelines that can process and curate massive datasets while maintaining quality and diversity

• Track record of designing controlled experiments that isolate the impact of architectural innovations and validate theoretical insights

• Experience optimizing models for real-world deployment, including knowledge of hardware constraints and efficiency techniques

• History of open-source contributions or research publications that have advanced the state of the art in speech/language AI

Ideal Researcher Profile

An ideal researcher, as evidenced by the recurring themes across these foundational papers, excels in five key areas: (1) Statistical & Mathematical Foundations, (2) Algorithmic Innovation & Implementation, (3) Data-Driven & Scalable Systems, (4) Hardware & Systems Understanding, and (5) Rigorous Experimental Design. Below is a synthesis of how each paper highlights these qualities, with references illustrating why they matter for building robust, impactful deep learning models.

1. Statistical & Mathematical Foundations

Mastery of Core Concepts

Many papers, like Scaling Laws for Neural Language Models and Neural Discrete Representation Learning (VQ-VAE), reflect the importance of power-law analyses, derivation of novel losses, or adaptation of fundamental equations (e.g., in VQ-VAE’s commitment loss or rectified flows in Scaling Rectified Flow Transformers). Such mathematical grounding clarifies why models converge or suffer collapse.

Combining Existing Theories in Novel Ways

Papers such as Moshi (combining text modeling, audio codecs, and hierarchical generative modeling) and Finite Scalar Quantization (FSQ’s adaptation of classic scalar quantization to replace vector-quantized representations) show how reusing but reimagining known techniques can yield breakthroughs. Many references (e.g., the structured state-space duality in Transformers are SSMs) underscore how unifying previously separate research lines can reveal powerful algorithmic or theoretical insights.

Logical Reasoning and Assumption Testing

Across all papers—particularly in the problem statements of Whisper or Rectified Flow Transformers—the authors present assumptions (e.g., “scaling data leads to zero-shot robustness” or “straight-line noise injection improves sample efficiency”) and systematically verify them with thorough empirical results. An ideal researcher similarly grounds new ideas in well-formed, testable hypotheses.

2. Algorithmic Innovation & Implementation

Creative Solutions to Known Bottlenecks

Each paper puts forth a unique algorithmic contribution—Rectified Flow Transformers redefines standard diffusion paths, FSQ proposes simpler scalar quantizations contrasted with VQ, phi-3 mini relies on curated data and blocksparse attention, and Mamba-2 merges SSM speed with attention concepts.

Turning Theory into Practice

Whether it’s the direct preference optimization (DPO) for alignment in phi-3 or the residual vector quantization in SoundStream, these works show that bridging design insights with implementable prototypes is essential.

Clear Impact Through Prototypes & Open-Source

Many references (Whisper, neural discrete representation learning, Mamba-2) highlight releasing code or pretrained models, enabling the broader community to replicate and build upon new methods. This premise of collaboration fosters faster progress.

3. Data-Driven & Scalable Systems

Emphasis on Large-Scale Data and Efficient Pipelines

Papers such as Robust Speech Recognition via Large-Scale Weak Supervision (Whisper) and BASE TTS demonstrate that collecting and processing hundreds of thousands of hours of real-world audio can unlock new capabilities in zero-shot or low-resource domains. Meanwhile, phi-3 Technical Report shows that filtering and curating data at scale (e.g., “data optimal regime”) can yield high performance even in smaller models.

Strategic Use of Data for Staged Training

A recurring strategy is to vary sources of data or the order of tasks. Whisper trains on multilingual tasks, BASE TTS uses subsets/stages for pretraining on speech tokens, and phi-3 deploys multiple training phases (web data, then synthetic data). This systematic approach to data underscores how an ideal researcher designs training curricula and data filtering protocols for maximum performance.

4. Hardware & Systems Understanding

Efficient Implementations at Scale

Many works illustrate how researchers tune architectures for modern accelerators: the In-Datacenter TPU paper exemplifies domain-specific hardware design for dense matrix multiplications, while phi-3 leverages blocksparse attention and custom Triton kernels to run advanced LLMs on resource-limited devices.

Real-Time & On-Device Constraints

SoundStream shows how to compress audio in real time on a smartphone CPU, demonstrating that knowledge of hardware constraints (latency, limited memory) drives design choices. Similarly, Moshi’s low-latency streaming TTS and phi-3-mini’s phone-based inference highlight that an ideal researcher must adapt algorithms to resource limits while maintaining robustness.

Architectural & Optimization Details

Papers like Mamba-2 in Transformers are SSMs and the In-Datacenter TPU work show how exploiting specialized matrix decomposition, custom memory hierarchies, or quantization approaches can lead to breakthroughs in speed or energy efficiency.

5. Rigorous Experimental Design

Controlled Comparisons & Ablations

Nearly all papers—Whisper, FSQ, Mamba-2, BASE TTS—use systematic ablations to isolate the impact of individual components (e.g., ablation on vector-quantization vs. scalar quantization in FSQ, or size of codebooks in VQ-VAEs). This approach reveals which design decisions truly matter.

Multifold Evaluation Metrics

From MUSHRA listening tests (SoundStream, BASE TTS) to FID in image synthesis (Scaling Rectified Flow Transformers, FSQ) to perplexity or zero-shot generalization in language (phi-3, Scaling Laws for Neural Language Models), the works demonstrate the value of comprehensive, carefully chosen metrics.

Stress Tests & Edge Cases

Whisper’s out-of-distribution speech benchmarks, SoundStream’s evaluation on speech + music, or Mamba-2’s performance on multi-query associative recall demonstrate the importance of specialized challenge sets. Researchers who craft or adopt rigorous benchmarks and “red-team” their models (as in phi-3 safety alignment) are better prepared to address real-world complexities.

Summary

Overall, an ideal researcher in deep learning consistently demonstrates:

• A solid grounding in theoretical and statistical principles

• A talent for proposing and validating new algorithmic solutions

• The capacity to orchestrate data pipelines that scale and reflect real-world diversity

• Awareness of hardware constraints and system-level trade-offs for efficiency

• Thorough and transparent experimental practices

These qualities surface across research on speech (Whisper, BASE TTS), language modeling (Scaling Laws, phi-3), specialized hardware (TPU, Transformers are SSMs), and new representation methods (VQ-VAE, FSQ, SoundStream). By balancing these attributes—rigorous math, innovative algorithms, large-scale data engineering, hardware-savvy optimizations, and reproducible experimentation—researchers can produce impactful, trustworthy advancements in foundational deep learning.

Foundational Papers

This job description was generated through analysis of the following papers:

• Robust Speech Recognition via Large-Scale Weak Supervision (arXiv:2212.04356)

• Moshi: a speech-text foundation model for real-time dialogue (arXiv:2410.00037)

• Scaling Rectified Flow Transformers for High-Resolution Image Synthesis (arXiv:2403.03206)

• Scaling Laws for Neural Language Models (arXiv:2001.08361)

• BASE TTS: Lessons from building a billion-parameter Text-to-Speech model on 100K hours of data (arXiv:2402.08093)

• In-Datacenter Performance Analysis of a Tensor Processing Unit (arXiv:1704.04760)

• Neural Discrete Representation Learning (arXiv:1711.00937)

• SoundStream: An End-to-End Neural Audio Codec (arXiv:2107.03312)

• Finite Scalar Quantization: VQ-VAE Made Simple (arXiv:2309.15505)

• Phi-3 Technical Report: A Highly Capable Language Model Locally on Your Phone (arXiv:2404.14219)

• Transformers are SSMs: Generalized Models and Efficient Algorithms Through Structured State Space Duality (arXiv:2405.21060)

Notice: We’re aware of individuals impersonating Deepgram recruiters. All legitimate Deepgram recruiting communication comes from an @ deepgram.com email address. If you’ve received a message claiming to be Deepgram, please forward it to careers@deepgram.com.

Company Overview

Deepgram is the leading platform underpinning the emerging trillion-dollar Voice AI economy, providing real-time APIs for speech-to-text (STT), text-to-speech (TTS), and building production-grade voice agents at scale. More than 200,000 developers and 1,300+ organizations build voice offerings that are ‘Powered by Deepgram’, including Twilio, Cloudflare, Sierra, Decagon, Vapi, Daily, Cresta, Granola, and Jack in the Box. Deepgram’s voice-native foundation models are accessed through cloud APIs or as self-hosted and on-premises software, with unmatched accuracy, low latency, and cost efficiency. Backed by a recent Series C led by leading global investors and strategic partners, Deepgram has processed over 50,000 years of audio and transcribed more than 1 trillion words. There is no organization in the world that understands voice better than Deepgram.

Company Operating Rhythm

At Deepgram, we expect an AI-first mindset—AI use and comfort aren’t optional, they’re core to how we operate, innovate, and measure performance.

Every team member who works at Deepgram is expected to actively use and experiment with advanced AI tools, and even build your own into your everyday work. We measure how effectively AI is applied to deliver results, and consistent, creative use of the latest AI capabilities is key to success here. Candidates should be comfortable adopting new models and modes quickly, integrating AI into their workflows, and continuously pushing the boundaries of what these technologies can do.

Additionally, we move at the pace of AI. Change is rapid, and you can expect your day-to-day work to evolve just as quickly. This may not be the right role if you’re not excited to experiment, adapt, think on your feet, and learn constantly, or if you’re seeking something highly prescriptive with a traditional 9-to-5.

The Opportunity

Voice is the most natural modality for human interaction with machines. However, current sequence modeling paradigms based on jointly scaling model and data cannot deliver voice AI capable of universal human interaction. The challenges are rooted in fundamental data problems posed by audio: real-world audio data is scarce and enormously diverse, spanning a vast space of voices, speaking styles, and acoustic conditions. Even if billions of hours of audio were accessible, its inherent high dimensionality creates computational and storage costs that make training and deployment prohibitively expensive at world scale. We believe that entirely new paradigms for audio AI are needed to overcome these challenges and make voice interaction accessible to everyone.

The Role

As a Member of the Research Staff, you will pioneer the development of Latent Space Models (LSMs), a new approach that aims to solve the fundamental data, scale, and cost challenges associated with building robust, contextualized voice AI. Your research will focus on solving one or more of the following problems:

• Build next-generation neural audio codecs that achieve extreme, low bit-rate compression and high fidelity reconstruction across a world-scale corpus of general audio.

• Pioneer steerable generative models that can synthesize the full diversity of human speech from the codec latent representation, from casual conversation to highly emotional expression to complex multi-speaker scenarios with environmental noise and overlapping speech.

• Develop embedding systems that cleanly factorize the codec latent space into interpretable dimensions of speaker, content, style, environment, and channel effects – enabling precise control over each aspect and the ability to massively amplify an existing seed dataset through “latent recombination”.

• Leverage latent recombination to generate synthetic audio data at previously impossible scales, unlocking joint model and data scaling paradigms for audio.  Endeavor to train multimodal speech-to-speech systems that can 1) understand any human irrespective of their demographics, state, or environment and 2) produce empathic, human-like responses that achieve conversational or task-oriented objectives.

• Design model architectures, training schemes, and inference algorithms that are adapted for hardware at the bare metal enabling cost efficient training on billion-hour datasets and powering real-time inference for hundreds of millions of concurrent conversations.

The Challenge

We are seeking researchers who:

• See “unsolved” problems as opportunities to pioneer entirely new approaches

• Can identify the one critical experiment that will validate or kill an idea in days, not months

• Have the vision to scale successful proofs-of-concept 100x

• Are obsessed with using AI to automate and amplify your own impact

If you find yourself energized rather than daunted by these expectations—if you’re already thinking about five ideas to try while reading this—you might be the researcher we need. This role demands obsession with the problems, creativity in approach, and relentless drive toward elegant, scalable solutions. The technical challenges are immense, but the potential impact is transformative.

It’s Important to Us That You Have

• Strong mathematical foundation in statistical learning theory, particularly in areas relevant to self-supervised and multimodal learning

• Deep expertise in foundation model architectures, with an understanding of how to scale training across multiple modalities

• Proven ability to bridge theory and practice—someone who can both derive novel mathematical formulations and implement them efficiently

• Demonstrated ability to build data pipelines that can process and curate massive datasets while maintaining quality and diversity

• Track record of designing controlled experiments that isolate the impact of architectural innovations and validate theoretical insights

• Experience optimizing models for real-world deployment, including knowledge of hardware constraints and efficiency techniques

• History of open-source contributions or research publications that have advanced the state of the art in speech/language AI

Ideal Researcher Profile

An ideal researcher, as evidenced by the recurring themes across these foundational papers, excels in five key areas: (1) Statistical & Mathematical Foundations, (2) Algorithmic Innovation & Implementation, (3) Data-Driven & Scalable Systems, (4) Hardware & Systems Understanding, and (5) Rigorous Experimental Design. Below is a synthesis of how each paper highlights these qualities, with references illustrating why they matter for building robust, impactful deep learning models.

1. Statistical & Mathematical Foundations

Mastery of Core Concepts

Many papers, like Scaling Laws for Neural Language Models and Neural Discrete Representation Learning (VQ-VAE), reflect the importance of power-law analyses, derivation of novel losses, or adaptation of fundamental equations (e.g., in VQ-VAE’s commitment loss or rectified flows in Scaling Rectified Flow Transformers). Such mathematical grounding clarifies why models converge or suffer collapse.

Combining Existing Theories in Novel Ways

Papers such as Moshi (combining text modeling, audio codecs, and hierarchical generative modeling) and Finite Scalar Quantization (FSQ’s adaptation of classic scalar quantization to replace vector-quantized representations) show how reusing but reimagining known techniques can yield breakthroughs. Many references (e.g., the structured state-space duality in Transformers are SSMs) underscore how unifying previously separate research lines can reveal powerful algorithmic or theoretical insights.

Logical Reasoning and Assumption Testing

Across all papers—particularly in the problem statements of Whisper or Rectified Flow Transformers—the authors present assumptions (e.g., “scaling data leads to zero-shot robustness” or “straight-line noise injection improves sample efficiency”) and systematically verify them with thorough empirical results. An ideal researcher similarly grounds new ideas in well-formed, testable hypotheses.

2. Algorithmic Innovation & Implementation

Creative Solutions to Known Bottlenecks

Each paper puts forth a unique algorithmic contribution—Rectified Flow Transformers redefines standard diffusion paths, FSQ proposes simpler scalar quantizations contrasted with VQ, phi-3 mini relies on curated data and blocksparse attention, and Mamba-2 merges SSM speed with attention concepts.

Turning Theory into Practice

Whether it’s the direct preference optimization (DPO) for alignment in phi-3 or the residual vector quantization in SoundStream, these works show that bridging design insights with implementable prototypes is essential.

Clear Impact Through Prototypes & Open-Source

Many references (Whisper, neural discrete representation learning, Mamba-2) highlight releasing code or pretrained models, enabling the broader community to replicate and build upon new methods. This premise of collaboration fosters faster progress.

3. Data-Driven & Scalable Systems

Emphasis on Large-Scale Data and Efficient Pipelines

Papers such as Robust Speech Recognition via Large-Scale Weak Supervision (Whisper) and BASE TTS demonstrate that collecting and processing hundreds of thousands of hours of real-world audio can unlock new capabilities in zero-shot or low-resource domains. Meanwhile, phi-3 Technical Report shows that filtering and curating data at scale (e.g., “data optimal regime”) can yield high performance even in smaller models.

Strategic Use of Data for Staged Training

A recurring strategy is to vary sources of data or the order of tasks. Whisper trains on multilingual tasks, BASE TTS uses subsets/stages for pretraining on speech tokens, and phi-3 deploys multiple training phases (web data, then synthetic data). This systematic approach to data underscores how an ideal researcher designs training curricula and data filtering protocols for maximum performance.

4. Hardware & Systems Understanding

Efficient Implementations at Scale

Many works illustrate how researchers tune architectures for modern accelerators: the In-Datacenter TPU paper exemplifies domain-specific hardware design for dense matrix multiplications, while phi-3 leverages blocksparse attention and custom Triton kernels to run advanced LLMs on resource-limited devices.

Real-Time & On-Device Constraints

SoundStream shows how to compress audio in real time on a smartphone CPU, demonstrating that knowledge of hardware constraints (latency, limited memory) drives design choices. Similarly, Moshi’s low-latency streaming TTS and phi-3-mini’s phone-based inference highlight that an ideal researcher must adapt algorithms to resource limits while maintaining robustness.

Architectural & Optimization Details

Papers like Mamba-2 in Transformers are SSMs and the In-Datacenter TPU work show how exploiting specialized matrix decomposition, custom memory hierarchies, or quantization approaches can lead to breakthroughs in speed or energy efficiency.

5. Rigorous Experimental Design

Controlled Comparisons & Ablations

Nearly all papers—Whisper, FSQ, Mamba-2, BASE TTS—use systematic ablations to isolate the impact of individual components (e.g., ablation on vector-quantization vs. scalar quantization in FSQ, or size of codebooks in VQ-VAEs). This approach reveals which design decisions truly matter.

Multifold Evaluation Metrics

From MUSHRA listening tests (SoundStream, BASE TTS) to FID in image synthesis (Scaling Rectified Flow Transformers, FSQ) to perplexity or zero-shot generalization in language (phi-3, Scaling Laws for Neural Language Models), the works demonstrate the value of comprehensive, carefully chosen metrics.

Stress Tests & Edge Cases

Whisper’s out-of-distribution speech benchmarks, SoundStream’s evaluation on speech + music, or Mamba-2’s performance on multi-query associative recall demonstrate the importance of specialized challenge sets. Researchers who craft or adopt rigorous benchmarks and “red-team” their models (as in phi-3 safety alignment) are better prepared to address real-world complexities.

Summary

Overall, an ideal researcher in deep learning consistently demonstrates:

• A solid grounding in theoretical and statistical principles

• A talent for proposing and validating new algorithmic solutions

• The capacity to orchestrate data pipelines that scale and reflect real-world diversity

• Awareness of hardware constraints and system-level trade-offs for efficiency

• Thorough and transparent experimental practices

These qualities surface across research on speech (Whisper, BASE TTS), language modeling (Scaling Laws, phi-3), specialized hardware (TPU, Transformers are SSMs), and new representation methods (VQ-VAE, FSQ, SoundStream). By balancing these attributes—rigorous math, innovative algorithms, large-scale data engineering, hardware-savvy optimizations, and reproducible experimentation—researchers can produce impactful, trustworthy advancements in foundational deep learning.

Foundational Papers

This job description was generated through analysis of the following papers:

• Robust Speech Recognition via Large-Scale Weak Supervision (arXiv:2212.04356)

• Moshi: a speech-text foundation model for real-time dialogue (arXiv:2410.00037)

• Scaling Rectified Flow Transformers for High-Resolution Image Synthesis (arXiv:2403.03206)

• Scaling Laws for Neural Language Models (arXiv:2001.08361)

• BASE TTS: Lessons from building a billion-parameter Text-to-Speech model on 100K hours of data (arXiv:2402.08093)

• In-Datacenter Performance Analysis of a Tensor Processing Unit (arXiv:1704.04760)

• Neural Discrete Representation Learning (arXiv:1711.00937)

• SoundStream: An End-to-End Neural Audio Codec (arXiv:2107.03312)

• Finite Scalar Quantization: VQ-VAE Made Simple (arXiv:2309.15505)

• Phi-3 Technical Report: A Highly Capable Language Model Locally on Your Phone (arXiv:2404.14219)

• Transformers are SSMs: Generalized Models and Efficient Algorithms Through Structured State Space Duality (arXiv:2405.21060)

Notice: We’re aware of individuals impersonating Deepgram recruiters. All legitimate Deepgram recruiting communication comes from an @ deepgram.com email address. If you’ve received a message claiming to be Deepgram, please forward it to careers@deepgram.com.

Company Overview

Deepgram is the leading platform underpinning the emerging trillion-dollar Voice AI economy, providing real-time APIs for speech-to-text (STT), text-to-speech (TTS), and building production-grade voice agents at scale. More than 200,000 developers and 1,300+ organizations build voice offerings that are ‘Powered by Deepgram’, including Twilio, Cloudflare, Sierra, Decagon, Vapi, Daily, Cresta, Granola, and Jack in the Box. Deepgram’s voice-native foundation models are accessed through cloud APIs or as self-hosted and on-premises software, with unmatched accuracy, low latency, and cost efficiency. Backed by a recent Series C led by leading global investors and strategic partners, Deepgram has processed over 50,000 years of audio and transcribed more than 1 trillion words. There is no organization in the world that understands voice better than Deepgram.

Company Operating Rhythm

At Deepgram, we expect an AI-first mindset—AI use and comfort aren’t optional, they’re core to how we operate, innovate, and measure performance.

Every team member who works at Deepgram is expected to actively use and experiment with advanced AI tools, and even build your own into your everyday work. We measure how effectively AI is applied to deliver results, and consistent, creative use of the latest AI capabilities is key to success here. Candidates should be comfortable adopting new models and modes quickly, integrating AI into their workflows, and continuously pushing the boundaries of what these technologies can do.

Additionally, we move at the pace of AI. Change is rapid, and you can expect your day-to-day work to evolve just as quickly. This may not be the right role if you’re not excited to experiment, adapt, think on your feet, and learn constantly, or if you’re seeking something highly prescriptive with a traditional 9-to-5.

The Opportunity

Voice is the most natural modality for human interaction with machines. However, current sequence modeling paradigms based on jointly scaling model and data cannot deliver voice AI capable of universal human interaction. The challenges are rooted in fundamental data problems posed by audio: real-world audio data is scarce and enormously diverse, spanning a vast space of voices, speaking styles, and acoustic conditions. Even if billions of hours of audio were accessible, its inherent high dimensionality creates computational and storage costs that make training and deployment prohibitively expensive at world scale. We believe that entirely new paradigms for audio AI are needed to overcome these challenges and make voice interaction accessible to everyone.

The Role

As a Member of the Research Staff, you will pioneer the development of Latent Space Models (LSMs), a new approach that aims to solve the fundamental data, scale, and cost challenges associated with building robust, contextualized voice AI. Your research will focus on solving one or more of the following problems:

• Build next-generation neural audio codecs that achieve extreme, low bit-rate compression and high fidelity reconstruction across a world-scale corpus of general audio.

• Pioneer steerable generative models that can synthesize the full diversity of human speech from the codec latent representation, from casual conversation to highly emotional expression to complex multi-speaker scenarios with environmental noise and overlapping speech.

• Develop embedding systems that cleanly factorize the codec latent space into interpretable dimensions of speaker, content, style, environment, and channel effects – enabling precise control over each aspect and the ability to massively amplify an existing seed dataset through “latent recombination”.

• Leverage latent recombination to generate synthetic audio data at previously impossible scales, unlocking joint model and data scaling paradigms for audio.  Endeavor to train multimodal speech-to-speech systems that can 1) understand any human irrespective of their demographics, state, or environment and 2) produce empathic, human-like responses that achieve conversational or task-oriented objectives.

• Design model architectures, training schemes, and inference algorithms that are adapted for hardware at the bare metal enabling cost efficient training on billion-hour datasets and powering real-time inference for hundreds of millions of concurrent conversations.

The Challenge

We are seeking researchers who:

• See “unsolved” problems as opportunities to pioneer entirely new approaches

• Can identify the one critical experiment that will validate or kill an idea in days, not months

• Have the vision to scale successful proofs-of-concept 100x

• Are obsessed with using AI to automate and amplify your own impact

If you find yourself energized rather than daunted by these expectations—if you’re already thinking about five ideas to try while reading this—you might be the researcher we need. This role demands obsession with the problems, creativity in approach, and relentless drive toward elegant, scalable solutions. The technical challenges are immense, but the potential impact is transformative.

It’s Important to Us That You Have

• Strong mathematical foundation in statistical learning theory, particularly in areas relevant to self-supervised and multimodal learning

• Deep expertise in foundation model architectures, with an understanding of how to scale training across multiple modalities

• Proven ability to bridge theory and practice—someone who can both derive novel mathematical formulations and implement them efficiently

• Demonstrated ability to build data pipelines that can process and curate massive datasets while maintaining quality and diversity

• Track record of designing controlled experiments that isolate the impact of architectural innovations and validate theoretical insights

• Experience optimizing models for real-world deployment, including knowledge of hardware constraints and efficiency techniques

• History of open-source contributions or research publications that have advanced the state of the art in speech/language AI

Ideal Researcher Profile

An ideal researcher, as evidenced by the recurring themes across these foundational papers, excels in five key areas: (1) Statistical & Mathematical Foundations, (2) Algorithmic Innovation & Implementation, (3) Data-Driven & Scalable Systems, (4) Hardware & Systems Understanding, and (5) Rigorous Experimental Design. Below is a synthesis of how each paper highlights these qualities, with references illustrating why they matter for building robust, impactful deep learning models.

1. Statistical & Mathematical Foundations

Mastery of Core Concepts

Many papers, like Scaling Laws for Neural Language Models and Neural Discrete Representation Learning (VQ-VAE), reflect the importance of power-law analyses, derivation of novel losses, or adaptation of fundamental equations (e.g., in VQ-VAE’s commitment loss or rectified flows in Scaling Rectified Flow Transformers). Such mathematical grounding clarifies why models converge or suffer collapse.

Combining Existing Theories in Novel Ways

Papers such as Moshi (combining text modeling, audio codecs, and hierarchical generative modeling) and Finite Scalar Quantization (FSQ’s adaptation of classic scalar quantization to replace vector-quantized representations) show how reusing but reimagining known techniques can yield breakthroughs. Many references (e.g., the structured state-space duality in Transformers are SSMs) underscore how unifying previously separate research lines can reveal powerful algorithmic or theoretical insights.

Logical Reasoning and Assumption Testing

Across all papers—particularly in the problem statements of Whisper or Rectified Flow Transformers—the authors present assumptions (e.g., “scaling data leads to zero-shot robustness” or “straight-line noise injection improves sample efficiency”) and systematically verify them with thorough empirical results. An ideal researcher similarly grounds new ideas in well-formed, testable hypotheses.

2. Algorithmic Innovation & Implementation

Creative Solutions to Known Bottlenecks

Each paper puts forth a unique algorithmic contribution—Rectified Flow Transformers redefines standard diffusion paths, FSQ proposes simpler scalar quantizations contrasted with VQ, phi-3 mini relies on curated data and blocksparse attention, and Mamba-2 merges SSM speed with attention concepts.

Turning Theory into Practice

Whether it’s the direct preference optimization (DPO) for alignment in phi-3 or the residual vector quantization in SoundStream, these works show that bridging design insights with implementable prototypes is essential.

Clear Impact Through Prototypes & Open-Source

Many references (Whisper, neural discrete representation learning, Mamba-2) highlight releasing code or pretrained models, enabling the broader community to replicate and build upon new methods. This premise of collaboration fosters faster progress.

3. Data-Driven & Scalable Systems

Emphasis on Large-Scale Data and Efficient Pipelines

Papers such as Robust Speech Recognition via Large-Scale Weak Supervision (Whisper) and BASE TTS demonstrate that collecting and processing hundreds of thousands of hours of real-world audio can unlock new capabilities in zero-shot or low-resource domains. Meanwhile, phi-3 Technical Report shows that filtering and curating data at scale (e.g., “data optimal regime”) can yield high performance even in smaller models.

Strategic Use of Data for Staged Training

A recurring strategy is to vary sources of data or the order of tasks. Whisper trains on multilingual tasks, BASE TTS uses subsets/stages for pretraining on speech tokens, and phi-3 deploys multiple training phases (web data, then synthetic data). This systematic approach to data underscores how an ideal researcher designs training curricula and data filtering protocols for maximum performance.

4. Hardware & Systems Understanding

Efficient Implementations at Scale

Many works illustrate how researchers tune architectures for modern accelerators: the In-Datacenter TPU paper exemplifies domain-specific hardware design for dense matrix multiplications, while phi-3 leverages blocksparse attention and custom Triton kernels to run advanced LLMs on resource-limited devices.

Real-Time & On-Device Constraints

SoundStream shows how to compress audio in real time on a smartphone CPU, demonstrating that knowledge of hardware constraints (latency, limited memory) drives design choices. Similarly, Moshi’s low-latency streaming TTS and phi-3-mini’s phone-based inference highlight that an ideal researcher must adapt algorithms to resource limits while maintaining robustness.

Architectural & Optimization Details

Papers like Mamba-2 in Transformers are SSMs and the In-Datacenter TPU work show how exploiting specialized matrix decomposition, custom memory hierarchies, or quantization approaches can lead to breakthroughs in speed or energy efficiency.

5. Rigorous Experimental Design

Controlled Comparisons & Ablations

Nearly all papers—Whisper, FSQ, Mamba-2, BASE TTS—use systematic ablations to isolate the impact of individual components (e.g., ablation on vector-quantization vs. scalar quantization in FSQ, or size of codebooks in VQ-VAEs). This approach reveals which design decisions truly matter.

Multifold Evaluation Metrics

From MUSHRA listening tests (SoundStream, BASE TTS) to FID in image synthesis (Scaling Rectified Flow Transformers, FSQ) to perplexity or zero-shot generalization in language (phi-3, Scaling Laws for Neural Language Models), the works demonstrate the value of comprehensive, carefully chosen metrics.

Stress Tests & Edge Cases

Whisper’s out-of-distribution speech benchmarks, SoundStream’s evaluation on speech + music, or Mamba-2’s performance on multi-query associative recall demonstrate the importance of specialized challenge sets. Researchers who craft or adopt rigorous benchmarks and “red-team” their models (as in phi-3 safety alignment) are better prepared to address real-world complexities.

Summary

Overall, an ideal researcher in deep learning consistently demonstrates:

• A solid grounding in theoretical and statistical principles

• A talent for proposing and validating new algorithmic solutions

• The capacity to orchestrate data pipelines that scale and reflect real-world diversity

• Awareness of hardware constraints and system-level trade-offs for efficiency

• Thorough and transparent experimental practices

These qualities surface across research on speech (Whisper, BASE TTS), language modeling (Scaling Laws, phi-3), specialized hardware (TPU, Transformers are SSMs), and new representation methods (VQ-VAE, FSQ, SoundStream). By balancing these attributes—rigorous math, innovative algorithms, large-scale data engineering, hardware-savvy optimizations, and reproducible experimentation—researchers can produce impactful, trustworthy advancements in foundational deep learning.

Foundational Papers

This job description was generated through analysis of the following papers:

• Robust Speech Recognition via Large-Scale Weak Supervision (arXiv:2212.04356)

• Moshi: a speech-text foundation model for real-time dialogue (arXiv:2410.00037)

• Scaling Rectified Flow Transformers for High-Resolution Image Synthesis (arXiv:2403.03206)

• Scaling Laws for Neural Language Models (arXiv:2001.08361)

• BASE TTS: Lessons from building a billion-parameter Text-to-Speech model on 100K hours of data (arXiv:2402.08093)

• In-Datacenter Performance Analysis of a Tensor Processing Unit (arXiv:1704.04760)

• Neural Discrete Representation Learning (arXiv:1711.00937)

• SoundStream: An End-to-End Neural Audio Codec (arXiv:2107.03312)

• Finite Scalar Quantization: VQ-VAE Made Simple (arXiv:2309.15505)

• Phi-3 Technical Report: A Highly Capable Language Model Locally on Your Phone (arXiv:2404.14219)

• Transformers are SSMs: Generalized Models and Efficient Algorithms Through Structured State Space Duality (arXiv:2405.21060)

Notice: We’re aware of individuals impersonating Deepgram recruiters. All legitimate Deepgram recruiting communication comes from an @ deepgram.com email address. If you’ve received a message claiming to be Deepgram, please forward it to careers@deepgram.com.

Company Overview

Deepgram is the leading platform underpinning the emerging trillion-dollar Voice AI economy, providing real-time APIs for speech-to-text (STT), text-to-speech (TTS), and building production-grade voice agents at scale. More than 200,000 developers and 1,300+ organizations build voice offerings that are ‘Powered by Deepgram’, including Twilio, Cloudflare, Sierra, Decagon, Vapi, Daily, Cresta, Granola, and Jack in the Box. Deepgram’s voice-native foundation models are accessed through cloud APIs or as self-hosted and on-premises software, with unmatched accuracy, low latency, and cost efficiency. Backed by a recent Series C led by leading global investors and strategic partners, Deepgram has processed over 50,000 years of audio and transcribed more than 1 trillion words. There is no organization in the world that understands voice better than Deepgram.

Company Operating Rhythm

At Deepgram, we expect an AI-first mindset—AI use and comfort aren’t optional, they’re core to how we operate, innovate, and measure performance.

Every team member who works at Deepgram is expected to actively use and experiment with advanced AI tools, and even build your own into your everyday work. We measure how effectively AI is applied to deliver results, and consistent, creative use of the latest AI capabilities is key to success here. Candidates should be comfortable adopting new models and modes quickly, integrating AI into their workflows, and continuously pushing the boundaries of what these technologies can do.

Additionally, we move at the pace of AI. Change is rapid, and you can expect your day-to-day work to evolve just as quickly. This may not be the right role if you’re not excited to experiment, adapt, think on your feet, and learn constantly, or if you’re seeking something highly prescriptive with a traditional 9-to-5.

The Opportunity

Voice is the most natural modality for human interaction with machines. However, current sequence modeling paradigms based on jointly scaling model and data cannot deliver voice AI capable of universal human interaction. The challenges are rooted in fundamental data problems posed by audio: real-world audio data is scarce and enormously diverse, spanning a vast space of voices, speaking styles, and acoustic conditions. Even if billions of hours of audio were accessible, its inherent high dimensionality creates computational and storage costs that make training and deployment prohibitively expensive at world scale. We believe that entirely new paradigms for audio AI are needed to overcome these challenges and make voice interaction accessible to everyone.

The Role

As a Member of the Research Staff, you will pioneer the development of Latent Space Models (LSMs), a new approach that aims to solve the fundamental data, scale, and cost challenges associated with building robust, contextualized voice AI. Your research will focus on solving one or more of the following problems:

• Build next-generation neural audio codecs that achieve extreme, low bit-rate compression and high fidelity reconstruction across a world-scale corpus of general audio.

• Pioneer steerable generative models that can synthesize the full diversity of human speech from the codec latent representation, from casual conversation to highly emotional expression to complex multi-speaker scenarios with environmental noise and overlapping speech.

• Develop embedding systems that cleanly factorize the codec latent space into interpretable dimensions of speaker, content, style, environment, and channel effects – enabling precise control over each aspect and the ability to massively amplify an existing seed dataset through “latent recombination”.

• Leverage latent recombination to generate synthetic audio data at previously impossible scales, unlocking joint model and data scaling paradigms for audio.  Endeavor to train multimodal speech-to-speech systems that can 1) understand any human irrespective of their demographics, state, or environment and 2) produce empathic, human-like responses that achieve conversational or task-oriented objectives.

• Design model architectures, training schemes, and inference algorithms that are adapted for hardware at the bare metal enabling cost efficient training on billion-hour datasets and powering real-time inference for hundreds of millions of concurrent conversations.

The Challenge

We are seeking researchers who:

• See “unsolved” problems as opportunities to pioneer entirely new approaches

• Can identify the one critical experiment that will validate or kill an idea in days, not months

• Have the vision to scale successful proofs-of-concept 100x

• Are obsessed with using AI to automate and amplify your own impact

If you find yourself energized rather than daunted by these expectations—if you’re already thinking about five ideas to try while reading this—you might be the researcher we need. This role demands obsession with the problems, creativity in approach, and relentless drive toward elegant, scalable solutions. The technical challenges are immense, but the potential impact is transformative.

It’s Important to Us That You Have

• Strong mathematical foundation in statistical learning theory, particularly in areas relevant to self-supervised and multimodal learning

• Deep expertise in foundation model architectures, with an understanding of how to scale training across multiple modalities

• Proven ability to bridge theory and practice—someone who can both derive novel mathematical formulations and implement them efficiently

• Demonstrated ability to build data pipelines that can process and curate massive datasets while maintaining quality and diversity

• Track record of designing controlled experiments that isolate the impact of architectural innovations and validate theoretical insights

• Experience optimizing models for real-world deployment, including knowledge of hardware constraints and efficiency techniques

• History of open-source contributions or research publications that have advanced the state of the art in speech/language AI

Ideal Researcher Profile

An ideal researcher, as evidenced by the recurring themes across these foundational papers, excels in five key areas: (1) Statistical & Mathematical Foundations, (2) Algorithmic Innovation & Implementation, (3) Data-Driven & Scalable Systems, (4) Hardware & Systems Understanding, and (5) Rigorous Experimental Design. Below is a synthesis of how each paper highlights these qualities, with references illustrating why they matter for building robust, impactful deep learning models.

1. Statistical & Mathematical Foundations

Mastery of Core Concepts

Many papers, like Scaling Laws for Neural Language Models and Neural Discrete Representation Learning (VQ-VAE), reflect the importance of power-law analyses, derivation of novel losses, or adaptation of fundamental equations (e.g., in VQ-VAE’s commitment loss or rectified flows in Scaling Rectified Flow Transformers). Such mathematical grounding clarifies why models converge or suffer collapse.

Combining Existing Theories in Novel Ways

Papers such as Moshi (combining text modeling, audio codecs, and hierarchical generative modeling) and Finite Scalar Quantization (FSQ’s adaptation of classic scalar quantization to replace vector-quantized representations) show how reusing but reimagining known techniques can yield breakthroughs. Many references (e.g., the structured state-space duality in Transformers are SSMs) underscore how unifying previously separate research lines can reveal powerful algorithmic or theoretical insights.

Logical Reasoning and Assumption Testing

Across all papers—particularly in the problem statements of Whisper or Rectified Flow Transformers—the authors present assumptions (e.g., “scaling data leads to zero-shot robustness” or “straight-line noise injection improves sample efficiency”) and systematically verify them with thorough empirical results. An ideal researcher similarly grounds new ideas in well-formed, testable hypotheses.

2. Algorithmic Innovation & Implementation

Creative Solutions to Known Bottlenecks

Each paper puts forth a unique algorithmic contribution—Rectified Flow Transformers redefines standard diffusion paths, FSQ proposes simpler scalar quantizations contrasted with VQ, phi-3 mini relies on curated data and blocksparse attention, and Mamba-2 merges SSM speed with attention concepts.

Turning Theory into Practice

Whether it’s the direct preference optimization (DPO) for alignment in phi-3 or the residual vector quantization in SoundStream, these works show that bridging design insights with implementable prototypes is essential.

Clear Impact Through Prototypes & Open-Source

Many references (Whisper, neural discrete representation learning, Mamba-2) highlight releasing code or pretrained models, enabling the broader community to replicate and build upon new methods. This premise of collaboration fosters faster progress.

3. Data-Driven & Scalable Systems

Emphasis on Large-Scale Data and Efficient Pipelines

Papers such as Robust Speech Recognition via Large-Scale Weak Supervision (Whisper) and BASE TTS demonstrate that collecting and processing hundreds of thousands of hours of real-world audio can unlock new capabilities in zero-shot or low-resource domains. Meanwhile, phi-3 Technical Report shows that filtering and curating data at scale (e.g., “data optimal regime”) can yield high performance even in smaller models.

Strategic Use of Data for Staged Training

A recurring strategy is to vary sources of data or the order of tasks. Whisper trains on multilingual tasks, BASE TTS uses subsets/stages for pretraining on speech tokens, and phi-3 deploys multiple training phases (web data, then synthetic data). This systematic approach to data underscores how an ideal researcher designs training curricula and data filtering protocols for maximum performance.

4. Hardware & Systems Understanding

Efficient Implementations at Scale

Many works illustrate how researchers tune architectures for modern accelerators: the In-Datacenter TPU paper exemplifies domain-specific hardware design for dense matrix multiplications, while phi-3 leverages blocksparse attention and custom Triton kernels to run advanced LLMs on resource-limited devices.

Real-Time & On-Device Constraints

SoundStream shows how to compress audio in real time on a smartphone CPU, demonstrating that knowledge of hardware constraints (latency, limited memory) drives design choices. Similarly, Moshi’s low-latency streaming TTS and phi-3-mini’s phone-based inference highlight that an ideal researcher must adapt algorithms to resource limits while maintaining robustness.

Architectural & Optimization Details

Papers like Mamba-2 in Transformers are SSMs and the In-Datacenter TPU work show how exploiting specialized matrix decomposition, custom memory hierarchies, or quantization approaches can lead to breakthroughs in speed or energy efficiency.

5. Rigorous Experimental Design

Controlled Comparisons & Ablations

Nearly all papers—Whisper, FSQ, Mamba-2, BASE TTS—use systematic ablations to isolate the impact of individual components (e.g., ablation on vector-quantization vs. scalar quantization in FSQ, or size of codebooks in VQ-VAEs). This approach reveals which design decisions truly matter.

Multifold Evaluation Metrics

From MUSHRA listening tests (SoundStream, BASE TTS) to FID in image synthesis (Scaling Rectified Flow Transformers, FSQ) to perplexity or zero-shot generalization in language (phi-3, Scaling Laws for Neural Language Models), the works demonstrate the value of comprehensive, carefully chosen metrics.

Stress Tests & Edge Cases

Whisper’s out-of-distribution speech benchmarks, SoundStream’s evaluation on speech + music, or Mamba-2’s performance on multi-query associative recall demonstrate the importance of specialized challenge sets. Researchers who craft or adopt rigorous benchmarks and “red-team” their models (as in phi-3 safety alignment) are better prepared to address real-world complexities.

Summary

Overall, an ideal researcher in deep learning consistently demonstrates:

• A solid grounding in theoretical and statistical principles

• A talent for proposing and validating new algorithmic solutions

• The capacity to orchestrate data pipelines that scale and reflect real-world diversity

• Awareness of hardware constraints and system-level trade-offs for efficiency

• Thorough and transparent experimental practices

These qualities surface across research on speech (Whisper, BASE TTS), language modeling (Scaling Laws, phi-3), specialized hardware (TPU, Transformers are SSMs), and new representation methods (VQ-VAE, FSQ, SoundStream). By balancing these attributes—rigorous math, innovative algorithms, large-scale data engineering, hardware-savvy optimizations, and reproducible experimentation—researchers can produce impactful, trustworthy advancements in foundational deep learning.

Foundational Papers

This job description was generated through analysis of the following papers:

• Robust Speech Recognition via Large-Scale Weak Supervision (arXiv:2212.04356)

• Moshi: a speech-text foundation model for real-time dialogue (arXiv:2410.00037)

• Scaling Rectified Flow Transformers for High-Resolution Image Synthesis (arXiv:2403.03206)

• Scaling Laws for Neural Language Models (arXiv:2001.08361)

• BASE TTS: Lessons from building a billion-parameter Text-to-Speech model on 100K hours of data (arXiv:2402.08093)

• In-Datacenter Performance Analysis of a Tensor Processing Unit (arXiv:1704.04760)

• Neural Discrete Representation Learning (arXiv:1711.00937)

• SoundStream: An End-to-End Neural Audio Codec (arXiv:2107.03312)

• Finite Scalar Quantization: VQ-VAE Made Simple (arXiv:2309.15505)

• Phi-3 Technical Report: A Highly Capable Language Model Locally on Your Phone (arXiv:2404.14219)

• Transformers are SSMs: Generalized Models and Efficient Algorithms Through Structured State Space Duality (arXiv:2405.21060)

Notice: We’re aware of individuals impersonating Deepgram recruiters. All legitimate Deepgram recruiting communication comes from an @ deepgram.com email address. If you’ve received a message claiming to be Deepgram, please forward it to careers@deepgram.com.

Company Overview

Deepgram is the leading platform underpinning the emerging trillion-dollar Voice AI economy, providing real-time APIs for speech-to-text (STT), text-to-speech (TTS), and building production-grade voice agents at scale. More than 200,000 developers and 1,300+ organizations build voice offerings that are ‘Powered by Deepgram’, including Twilio, Cloudflare, Sierra, Decagon, Vapi, Daily, Cresta, Granola, and Jack in the Box. Deepgram’s voice-native foundation models are accessed through cloud APIs or as self-hosted and on-premises software, with unmatched accuracy, low latency, and cost efficiency. Backed by a recent Series C led by leading global investors and strategic partners, Deepgram has processed over 50,000 years of audio and transcribed more than 1 trillion words. There is no organization in the world that understands voice better than Deepgram.

Company Operating Rhythm

At Deepgram, we expect an AI-first mindset—AI use and comfort aren’t optional, they’re core to how we operate, innovate, and measure performance.

Every team member who works at Deepgram is expected to actively use and experiment with advanced AI tools, and even build your own into your everyday work. We measure how effectively AI is applied to deliver results, and consistent, creative use of the latest AI capabilities is key to success here. Candidates should be comfortable adopting new models and modes quickly, integrating AI into their workflows, and continuously pushing the boundaries of what these technologies can do.

Additionally, we move at the pace of AI. Change is rapid, and you can expect your day-to-day work to evolve just as quickly. This may not be the right role if you’re not excited to experiment, adapt, think on your feet, and learn constantly, or if you’re seeking something highly prescriptive with a traditional 9-to-5.

The Opportunity

Voice is the most natural modality for human interaction with machines. However, current sequence modeling paradigms based on jointly scaling model and data cannot deliver voice AI capable of universal human interaction. The challenges are rooted in fundamental data problems posed by audio: real-world audio data is scarce and enormously diverse, spanning a vast space of voices, speaking styles, and acoustic conditions. Even if billions of hours of audio were accessible, its inherent high dimensionality creates computational and storage costs that make training and deployment prohibitively expensive at world scale. We believe that entirely new paradigms for audio AI are needed to overcome these challenges and make voice interaction accessible to everyone.

The Role

As a Member of the Research Staff, you will pioneer the development of Latent Space Models (LSMs), a new approach that aims to solve the fundamental data, scale, and cost challenges associated with building robust, contextualized voice AI. Your research will focus on solving one or more of the following problems:

• Build next-generation neural audio codecs that achieve extreme, low bit-rate compression and high fidelity reconstruction across a world-scale corpus of general audio.

• Pioneer steerable generative models that can synthesize the full diversity of human speech from the codec latent representation, from casual conversation to highly emotional expression to complex multi-speaker scenarios with environmental noise and overlapping speech.

• Develop embedding systems that cleanly factorize the codec latent space into interpretable dimensions of speaker, content, style, environment, and channel effects – enabling precise control over each aspect and the ability to massively amplify an existing seed dataset through “latent recombination”.

• Leverage latent recombination to generate synthetic audio data at previously impossible scales, unlocking joint model and data scaling paradigms for audio.  Endeavor to train multimodal speech-to-speech systems that can 1) understand any human irrespective of their demographics, state, or environment and 2) produce empathic, human-like responses that achieve conversational or task-oriented objectives.

• Design model architectures, training schemes, and inference algorithms that are adapted for hardware at the bare metal enabling cost efficient training on billion-hour datasets and powering real-time inference for hundreds of millions of concurrent conversations.

The Challenge

We are seeking researchers who:

• See “unsolved” problems as opportunities to pioneer entirely new approaches

• Can identify the one critical experiment that will validate or kill an idea in days, not months

• Have the vision to scale successful proofs-of-concept 100x

• Are obsessed with using AI to automate and amplify your own impact

If you find yourself energized rather than daunted by these expectations—if you’re already thinking about five ideas to try while reading this—you might be the researcher we need. This role demands obsession with the problems, creativity in approach, and relentless drive toward elegant, scalable solutions. The technical challenges are immense, but the potential impact is transformative.

It’s Important to Us That You Have

• Strong mathematical foundation in statistical learning theory, particularly in areas relevant to self-supervised and multimodal learning

• Deep expertise in foundation model architectures, with an understanding of how to scale training across multiple modalities

• Proven ability to bridge theory and practice—someone who can both derive novel mathematical formulations and implement them efficiently

• Demonstrated ability to build data pipelines that can process and curate massive datasets while maintaining quality and diversity

• Track record of designing controlled experiments that isolate the impact of architectural innovations and validate theoretical insights

• Experience optimizing models for real-world deployment, including knowledge of hardware constraints and efficiency techniques

• History of open-source contributions or research publications that have advanced the state of the art in speech/language AI

Ideal Researcher Profile

An ideal researcher, as evidenced by the recurring themes across these foundational papers, excels in five key areas: (1) Statistical & Mathematical Foundations, (2) Algorithmic Innovation & Implementation, (3) Data-Driven & Scalable Systems, (4) Hardware & Systems Understanding, and (5) Rigorous Experimental Design. Below is a synthesis of how each paper highlights these qualities, with references illustrating why they matter for building robust, impactful deep learning models.

1. Statistical & Mathematical Foundations

Mastery of Core Concepts

Many papers, like Scaling Laws for Neural Language Models and Neural Discrete Representation Learning (VQ-VAE), reflect the importance of power-law analyses, derivation of novel losses, or adaptation of fundamental equations (e.g., in VQ-VAE’s commitment loss or rectified flows in Scaling Rectified Flow Transformers). Such mathematical grounding clarifies why models converge or suffer collapse.

Combining Existing Theories in Novel Ways

Papers such as Moshi (combining text modeling, audio codecs, and hierarchical generative modeling) and Finite Scalar Quantization (FSQ’s adaptation of classic scalar quantization to replace vector-quantized representations) show how reusing but reimagining known techniques can yield breakthroughs. Many references (e.g., the structured state-space duality in Transformers are SSMs) underscore how unifying previously separate research lines can reveal powerful algorithmic or theoretical insights.

Logical Reasoning and Assumption Testing

Across all papers—particularly in the problem statements of Whisper or Rectified Flow Transformers—the authors present assumptions (e.g., “scaling data leads to zero-shot robustness” or “straight-line noise injection improves sample efficiency”) and systematically verify them with thorough empirical results. An ideal researcher similarly grounds new ideas in well-formed, testable hypotheses.

2. Algorithmic Innovation & Implementation

Creative Solutions to Known Bottlenecks

Each paper puts forth a unique algorithmic contribution—Rectified Flow Transformers redefines standard diffusion paths, FSQ proposes simpler scalar quantizations contrasted with VQ, phi-3 mini relies on curated data and blocksparse attention, and Mamba-2 merges SSM speed with attention concepts.

Turning Theory into Practice

Whether it’s the direct preference optimization (DPO) for alignment in phi-3 or the residual vector quantization in SoundStream, these works show that bridging design insights with implementable prototypes is essential.

Clear Impact Through Prototypes & Open-Source

Many references (Whisper, neural discrete representation learning, Mamba-2) highlight releasing code or pretrained models, enabling the broader community to replicate and build upon new methods. This premise of collaboration fosters faster progress.

3. Data-Driven & Scalable Systems

Emphasis on Large-Scale Data and Efficient Pipelines

Papers such as Robust Speech Recognition via Large-Scale Weak Supervision (Whisper) and BASE TTS demonstrate that collecting and processing hundreds of thousands of hours of real-world audio can unlock new capabilities in zero-shot or low-resource domains. Meanwhile, phi-3 Technical Report shows that filtering and curating data at scale (e.g., “data optimal regime”) can yield high performance even in smaller models.

Strategic Use of Data for Staged Training

A recurring strategy is to vary sources of data or the order of tasks. Whisper trains on multilingual tasks, BASE TTS uses subsets/stages for pretraining on speech tokens, and phi-3 deploys multiple training phases (web data, then synthetic data). This systematic approach to data underscores how an ideal researcher designs training curricula and data filtering protocols for maximum performance.

4. Hardware & Systems Understanding

Efficient Implementations at Scale

Many works illustrate how researchers tune architectures for modern accelerators: the In-Datacenter TPU paper exemplifies domain-specific hardware design for dense matrix multiplications, while phi-3 leverages blocksparse attention and custom Triton kernels to run advanced LLMs on resource-limited devices.

Real-Time & On-Device Constraints

SoundStream shows how to compress audio in real time on a smartphone CPU, demonstrating that knowledge of hardware constraints (latency, limited memory) drives design choices. Similarly, Moshi’s low-latency streaming TTS and phi-3-mini’s phone-based inference highlight that an ideal researcher must adapt algorithms to resource limits while maintaining robustness.

Architectural & Optimization Details

Papers like Mamba-2 in Transformers are SSMs and the In-Datacenter TPU work show how exploiting specialized matrix decomposition, custom memory hierarchies, or quantization approaches can lead to breakthroughs in speed or energy efficiency.

5. Rigorous Experimental Design

Controlled Comparisons & Ablations

Nearly all papers—Whisper, FSQ, Mamba-2, BASE TTS—use systematic ablations to isolate the impact of individual components (e.g., ablation on vector-quantization vs. scalar quantization in FSQ, or size of codebooks in VQ-VAEs). This approach reveals which design decisions truly matter.

Multifold Evaluation Metrics

From MUSHRA listening tests (SoundStream, BASE TTS) to FID in image synthesis (Scaling Rectified Flow Transformers, FSQ) to perplexity or zero-shot generalization in language (phi-3, Scaling Laws for Neural Language Models), the works demonstrate the value of comprehensive, carefully chosen metrics.

Stress Tests & Edge Cases

Whisper’s out-of-distribution speech benchmarks, SoundStream’s evaluation on speech + music, or Mamba-2’s performance on multi-query associative recall demonstrate the importance of specialized challenge sets. Researchers who craft or adopt rigorous benchmarks and “red-team” their models (as in phi-3 safety alignment) are better prepared to address real-world complexities.

Summary

Overall, an ideal researcher in deep learning consistently demonstrates:

• A solid grounding in theoretical and statistical principles

• A talent for proposing and validating new algorithmic solutions

• The capacity to orchestrate data pipelines that scale and reflect real-world diversity

• Awareness of hardware constraints and system-level trade-offs for efficiency

• Thorough and transparent experimental practices

These qualities surface across research on speech (Whisper, BASE TTS), language modeling (Scaling Laws, phi-3), specialized hardware (TPU, Transformers are SSMs), and new representation methods (VQ-VAE, FSQ, SoundStream). By balancing these attributes—rigorous math, innovative algorithms, large-scale data engineering, hardware-savvy optimizations, and reproducible experimentation—researchers can produce impactful, trustworthy advancements in foundational deep learning.

Foundational Papers

This job description was generated through analysis of the following papers:

• Robust Speech Recognition via Large-Scale Weak Supervision (arXiv:2212.04356)

• Moshi: a speech-text foundation model for real-time dialogue (arXiv:2410.00037)

• Scaling Rectified Flow Transformers for High-Resolution Image Synthesis (arXiv:2403.03206)

• Scaling Laws for Neural Language Models (arXiv:2001.08361)

• BASE TTS: Lessons from building a billion-parameter Text-to-Speech model on 100K hours of data (arXiv:2402.08093)

• In-Datacenter Performance Analysis of a Tensor Processing Unit (arXiv:1704.04760)

• Neural Discrete Representation Learning (arXiv:1711.00937)

• SoundStream: An End-to-End Neural Audio Codec (arXiv:2107.03312)

• Finite Scalar Quantization: VQ-VAE Made Simple (arXiv:2309.15505)

• Phi-3 Technical Report: A Highly Capable Language Model Locally on Your Phone (arXiv:2404.14219)

• Transformers are SSMs: Generalized Models and Efficient Algorithms Through Structured State Space Duality (arXiv:2405.21060)

Notice: We’re aware of individuals impersonating Deepgram recruiters. All legitimate Deepgram recruiting communication comes from an @ deepgram.com email address. If you’ve received a message claiming to be Deepgram, please forward it to careers@deepgram.com.

Company Overview

Deepgram is the leading platform underpinning the emerging trillion-dollar Voice AI economy, providing real-time APIs for speech-to-text (STT), text-to-speech (TTS), and building production-grade voice agents at scale. More than 200,000 developers and 1,300+ organizations build voice offerings that are ‘Powered by Deepgram’, including Twilio, Cloudflare, Sierra, Decagon, Vapi, Daily, Cresta, Granola, and Jack in the Box. Deepgram’s voice-native foundation models are accessed through cloud APIs or as self-hosted and on-premises software, with unmatched accuracy, low latency, and cost efficiency. Backed by a recent Series C led by leading global investors and strategic partners, Deepgram has processed over 50,000 years of audio and transcribed more than 1 trillion words. There is no organization in the world that understands voice better than Deepgram.

Company Operating Rhythm

At Deepgram, we expect an AI-first mindset—AI use and comfort aren’t optional, they’re core to how we operate, innovate, and measure performance.

Every team member who works at Deepgram is expected to actively use and experiment with advanced AI tools, and even build your own into your everyday work. We measure how effectively AI is applied to deliver results, and consistent, creative use of the latest AI capabilities is key to success here. Candidates should be comfortable adopting new models and modes quickly, integrating AI into their workflows, and continuously pushing the boundaries of what these technologies can do.

Additionally, we move at the pace of AI. Change is rapid, and you can expect your day-to-day work to evolve just as quickly. This may not be the right role if you’re not excited to experiment, adapt, think on your feet, and learn constantly, or if you’re seeking something highly prescriptive with a traditional 9-to-5.

The Opportunity

Voice is the most natural modality for human interaction with machines. However, current sequence modeling paradigms based on jointly scaling model and data cannot deliver voice AI capable of universal human interaction. The challenges are rooted in fundamental data problems posed by audio: real-world audio data is scarce and enormously diverse, spanning a vast space of voices, speaking styles, and acoustic conditions. Even if billions of hours of audio were accessible, its inherent high dimensionality creates computational and storage costs that make training and deployment prohibitively expensive at world scale. We believe that entirely new paradigms for audio AI are needed to overcome these challenges and make voice interaction accessible to everyone.

The Role

As a Member of the Research Staff, you will pioneer the development of Latent Space Models (LSMs), a new approach that aims to solve the fundamental data, scale, and cost challenges associated with building robust, contextualized voice AI. Your research will focus on solving one or more of the following problems:

• Build next-generation neural audio codecs that achieve extreme, low bit-rate compression and high fidelity reconstruction across a world-scale corpus of general audio.

• Pioneer steerable generative models that can synthesize the full diversity of human speech from the codec latent representation, from casual conversation to highly emotional expression to complex multi-speaker scenarios with environmental noise and overlapping speech.

• Develop embedding systems that cleanly factorize the codec latent space into interpretable dimensions of speaker, content, style, environment, and channel effects – enabling precise control over each aspect and the ability to massively amplify an existing seed dataset through “latent recombination”.

• Leverage latent recombination to generate synthetic audio data at previously impossible scales, unlocking joint model and data scaling paradigms for audio.  Endeavor to train multimodal speech-to-speech systems that can 1) understand any human irrespective of their demographics, state, or environment and 2) produce empathic, human-like responses that achieve conversational or task-oriented objectives.

• Design model architectures, training schemes, and inference algorithms that are adapted for hardware at the bare metal enabling cost efficient training on billion-hour datasets and powering real-time inference for hundreds of millions of concurrent conversations.

The Challenge

We are seeking researchers who:

• See “unsolved” problems as opportunities to pioneer entirely new approaches

• Can identify the one critical experiment that will validate or kill an idea in days, not months

• Have the vision to scale successful proofs-of-concept 100x

• Are obsessed with using AI to automate and amplify your own impact

If you find yourself energized rather than daunted by these expectations—if you’re already thinking about five ideas to try while reading this—you might be the researcher we need. This role demands obsession with the problems, creativity in approach, and relentless drive toward elegant, scalable solutions. The technical challenges are immense, but the potential impact is transformative.

It’s Important to Us That You Have

• Strong mathematical foundation in statistical learning theory, particularly in areas relevant to self-supervised and multimodal learning

• Deep expertise in foundation model architectures, with an understanding of how to scale training across multiple modalities

• Proven ability to bridge theory and practice—someone who can both derive novel mathematical formulations and implement them efficiently

• Demonstrated ability to build data pipelines that can process and curate massive datasets while maintaining quality and diversity

• Track record of designing controlled experiments that isolate the impact of architectural innovations and validate theoretical insights

• Experience optimizing models for real-world deployment, including knowledge of hardware constraints and efficiency techniques

• History of open-source contributions or research publications that have advanced the state of the art in speech/language AI

Ideal Researcher Profile

An ideal researcher, as evidenced by the recurring themes across these foundational papers, excels in five key areas: (1) Statistical & Mathematical Foundations, (2) Algorithmic Innovation & Implementation, (3) Data-Driven & Scalable Systems, (4) Hardware & Systems Understanding, and (5) Rigorous Experimental Design. Below is a synthesis of how each paper highlights these qualities, with references illustrating why they matter for building robust, impactful deep learning models.

1. Statistical & Mathematical Foundations

Mastery of Core Concepts

Many papers, like Scaling Laws for Neural Language Models and Neural Discrete Representation Learning (VQ-VAE), reflect the importance of power-law analyses, derivation of novel losses, or adaptation of fundamental equations (e.g., in VQ-VAE’s commitment loss or rectified flows in Scaling Rectified Flow Transformers). Such mathematical grounding clarifies why models converge or suffer collapse.

Combining Existing Theories in Novel Ways

Papers such as Moshi (combining text modeling, audio codecs, and hierarchical generative modeling) and Finite Scalar Quantization (FSQ’s adaptation of classic scalar quantization to replace vector-quantized representations) show how reusing but reimagining known techniques can yield breakthroughs. Many references (e.g., the structured state-space duality in Transformers are SSMs) underscore how unifying previously separate research lines can reveal powerful algorithmic or theoretical insights.

Logical Reasoning and Assumption Testing

Across all papers—particularly in the problem statements of Whisper or Rectified Flow Transformers—the authors present assumptions (e.g., “scaling data leads to zero-shot robustness” or “straight-line noise injection improves sample efficiency”) and systematically verify them with thorough empirical results. An ideal researcher similarly grounds new ideas in well-formed, testable hypotheses.

2. Algorithmic Innovation & Implementation

Creative Solutions to Known Bottlenecks

Each paper puts forth a unique algorithmic contribution—Rectified Flow Transformers redefines standard diffusion paths, FSQ proposes simpler scalar quantizations contrasted with VQ, phi-3 mini relies on curated data and blocksparse attention, and Mamba-2 merges SSM speed with attention concepts.

Turning Theory into Practice

Whether it’s the direct preference optimization (DPO) for alignment in phi-3 or the residual vector quantization in SoundStream, these works show that bridging design insights with implementable prototypes is essential.

Clear Impact Through Prototypes & Open-Source

Many references (Whisper, neural discrete representation learning, Mamba-2) highlight releasing code or pretrained models, enabling the broader community to replicate and build upon new methods. This premise of collaboration fosters faster progress.

3. Data-Driven & Scalable Systems

Emphasis on Large-Scale Data and Efficient Pipelines

Papers such as Robust Speech Recognition via Large-Scale Weak Supervision (Whisper) and BASE TTS demonstrate that collecting and processing hundreds of thousands of hours of real-world audio can unlock new capabilities in zero-shot or low-resource domains. Meanwhile, phi-3 Technical Report shows that filtering and curating data at scale (e.g., “data optimal regime”) can yield high performance even in smaller models.

Strategic Use of Data for Staged Training

A recurring strategy is to vary sources of data or the order of tasks. Whisper trains on multilingual tasks, BASE TTS uses subsets/stages for pretraining on speech tokens, and phi-3 deploys multiple training phases (web data, then synthetic data). This systematic approach to data underscores how an ideal researcher designs training curricula and data filtering protocols for maximum performance.

4. Hardware & Systems Understanding

Efficient Implementations at Scale

Many works illustrate how researchers tune architectures for modern accelerators: the In-Datacenter TPU paper exemplifies domain-specific hardware design for dense matrix multiplications, while phi-3 leverages blocksparse attention and custom Triton kernels to run advanced LLMs on resource-limited devices.

Real-Time & On-Device Constraints

SoundStream shows how to compress audio in real time on a smartphone CPU, demonstrating that knowledge of hardware constraints (latency, limited memory) drives design choices. Similarly, Moshi’s low-latency streaming TTS and phi-3-mini’s phone-based inference highlight that an ideal researcher must adapt algorithms to resource limits while maintaining robustness.

Architectural & Optimization Details

Papers like Mamba-2 in Transformers are SSMs and the In-Datacenter TPU work show how exploiting specialized matrix decomposition, custom memory hierarchies, or quantization approaches can lead to breakthroughs in speed or energy efficiency.

5. Rigorous Experimental Design

Controlled Comparisons & Ablations

Nearly all papers—Whisper, FSQ, Mamba-2, BASE TTS—use systematic ablations to isolate the impact of individual components (e.g., ablation on vector-quantization vs. scalar quantization in FSQ, or size of codebooks in VQ-VAEs). This approach reveals which design decisions truly matter.

Multifold Evaluation Metrics

From MUSHRA listening tests (SoundStream, BASE TTS) to FID in image synthesis (Scaling Rectified Flow Transformers, FSQ) to perplexity or zero-shot generalization in language (phi-3, Scaling Laws for Neural Language Models), the works demonstrate the value of comprehensive, carefully chosen metrics.

Stress Tests & Edge Cases

Whisper’s out-of-distribution speech benchmarks, SoundStream’s evaluation on speech + music, or Mamba-2’s performance on multi-query associative recall demonstrate the importance of specialized challenge sets. Researchers who craft or adopt rigorous benchmarks and “red-team” their models (as in phi-3 safety alignment) are better prepared to address real-world complexities.

Summary

Overall, an ideal researcher in deep learning consistently demonstrates:

• A solid grounding in theoretical and statistical principles

• A talent for proposing and validating new algorithmic solutions

• The capacity to orchestrate data pipelines that scale and reflect real-world diversity

• Awareness of hardware constraints and system-level trade-offs for efficiency

• Thorough and transparent experimental practices

These qualities surface across research on speech (Whisper, BASE TTS), language modeling (Scaling Laws, phi-3), specialized hardware (TPU, Transformers are SSMs), and new representation methods (VQ-VAE, FSQ, SoundStream). By balancing these attributes—rigorous math, innovative algorithms, large-scale data engineering, hardware-savvy optimizations, and reproducible experimentation—researchers can produce impactful, trustworthy advancements in foundational deep learning.

Foundational Papers

This job description was generated through analysis of the following papers:

• Robust Speech Recognition via Large-Scale Weak Supervision (arXiv:2212.04356)

• Moshi: a speech-text foundation model for real-time dialogue (arXiv:2410.00037)

• Scaling Rectified Flow Transformers for High-Resolution Image Synthesis (arXiv:2403.03206)

• Scaling Laws for Neural Language Models (arXiv:2001.08361)

• BASE TTS: Lessons from building a billion-parameter Text-to-Speech model on 100K hours of data (arXiv:2402.08093)

• In-Datacenter Performance Analysis of a Tensor Processing Unit (arXiv:1704.04760)

• Neural Discrete Representation Learning (arXiv:1711.00937)

• SoundStream: An End-to-End Neural Audio Codec (arXiv:2107.03312)

• Finite Scalar Quantization: VQ-VAE Made Simple (arXiv:2309.15505)

• Phi-3 Technical Report: A Highly Capable Language Model Locally on Your Phone (arXiv:2404.14219)

• Transformers are SSMs: Generalized Models and Efficient Algorithms Through Structured State Space Duality (arXiv:2405.21060)

Notice: We’re aware of individuals impersonating Deepgram recruiters. All legitimate Deepgram recruiting communication comes from an @ deepgram.com email address. If you’ve received a message claiming to be Deepgram, please forward it to careers@deepgram.com.

Build a Safer World. TRM Labs provides AI-powered intelligence solutions that help public and private sector agencies investigate and disrupt crime. TRM’s platforms enable investigators to trace illicit activity, build...

Job Title: Senior Archaeologist

Job Location: Remote

Job Code: SA-2026

Job Link: Senior Archaeologist - Southeastern Archaeological Research, LLC. “SEARCH”

SEARCH is seeking a Senior Archaeologist to support various projects in the US. The position can be remote or based out of one of our offices is Orlando, Jacksonville, Pensacola, or New Orleans.

Successful candidates will have a background in cultural resource management. The candidate should have excellent technical writing skills and the ability to produce reports for review by subject matter experts on behalf of federal, state, and tribal agencies under Section 106 of the National Historic Preservation Act and the National Environmental Policy Act, among other historic preservation laws/regulations.

• Use experience, education from a degreed program, and professional judgment to identify and target locations of enhanced site probability based on cultural, geological, and other environmental features
• Administer the project health and safety plan, participate in office and field training, and offer recommendations for improvement to health and safety manager
• ·Have a thorough understanding of the research design and ensure its implementation
• Based upon education, field observations, and scientific judgement, revise field methodology to meet overall project goals, in consultation with project leadership
• Perform pre-field background research, prepare maps, input data into electronic devices, and prepare project summary for field staff
• Provide leadership to field teams conducting archaeological monitoring, survey, testing, and excavation in support of Phase I, II, and III studies in accordance with the research design to identify, record, and evaluate archaeological resources
• Recognize archaeological signatures and apply reasoning to evaluate the extent and integrity of resources
• Record accurate, organized, and clear field notes, maps, drawings, logs, forms, tables, databases, artifact bags, box labels, and other project documentation
• Collect and inventory artifacts, ecofacts, and samples in accordance with the research design, and keep an accurate field specimen log
• Identify potential human and faunal remains, and treat any suspected human remains respectfully and in a manner consistent with the research design and applicable state and federal guidelines and laws
• Perform quality control of field data (hand-written and electronic) daily to ensure information is accurate and complete
• Use electronic and traditional navigational equipment (GPS, Field Maps, compass and transit) to determine project limits, record fieldwork, document findings, and communicate progress
• Photo-document artifacts, features, excavation units, land use, landscapes, unique environmental features, historic structures, archaeological sites, unique environmental features, specialized equipment or technique usage, and work-in-progress (keep an accurate photo log)
• Use or review GIS data to organize, interpret, and document field research
• Under supervision of the principal investigator, identify, record, and prepare preliminary evaluations of archaeological resources in accordance with state and federal laws, particularly the National Register of Historical Places (NRHP)
• Provide daily updates on project progress to the principal investigator or project manager
• Meet with, coordinate, and collaborate with landowners, land agents, clients, other field research teams, and visiting regulatory agency staff
• Interview informants to gather information on history and use of property
• Prepare thorough, accurate, professional, high-quality technical reports, including writing tasks, graphics, report production, and quality control
• Assist the principal investigator or project manager with other research, project management, and administrative tasks
• Prepare research permit applications for review and submittal by the principal investigator
• Under the supervision of a principal investigator or project manager, draft proposals and cost estimates
• Procure and maintain field equipment, technology, vehicles, and supplies in a clean, working, and safe manner and report any malfunctions, shortages, and risks to principal investigator or project manager
• Complete timesheets and other administrative or project requirements accurately and on schedule
• Understand and adhere to the SEARCH Brand Guidelines, SEARCH Style Guide, SEARCH Fieldwork Guide, and the SEARCH Employee Manual, and encourage others to do the same

Application Requirements

• Cover letter
• CV/resume
• Minimum of three professional references
• Three writing samples (if a collaborative report, identify sections authored). Writing samples should be sent to HR@searchinc.com. Include the Job Posting code (SA-2026) in the subject line of your email.

Other Important Need-to-Know Information

• Applicants must be eligible to be employed in the United States.

• Applicants must have a valid Driver’s License and reliable transportation.

• SEARCH offers a competitive total compensation package with strong health and wellness and retirement plans.

• Pay is commensurate with education and experience.

• SEARCH is strongly committed to creating a diverse and inclusive environment where a variety of backgrounds, cultures, orientations, ideas, and talents can flourish.

• SEARCH is an equal opportunity employer, and all qualified applicants will receive consideration for employment without regard to race, color, religion, sex, national origin, disability status, protected veteran status, or any other characteristic protected by law.

• SEARCH is an Affirmative Action Employer and a Drug-Free Workplace.

• Employment offers are contingent upon satisfactory verification of criminal, education, driving, and/or employment background checks, as well as passing results from a pre‑employment drug screen.

About Us:

SEARCH is a global leader in archaeology, maritime archaeology, and the full spectrum of cultural heritage services. Our staff are industry leaders and subject matter experts who support commercial and government clients worldwide. Since 1993, we have completed over 5,700 projects nationwide and in over 40 countries across seven continents and four oceans. By integrating science, technology, and creativity, we harness the power of the past to advance the projects, places, and people we serve.

• Bachelor’s degree in anthropology, archaeology, or a closely related field

• Completion of an accredited archaeological field school, six months of field experience, or both

• Minimum five years of professional archaeological field experience encompassing some supervisory experience

• Valid driver’s license. good driving record, access to reliable transportation

• Ability to obtain government security clearance

SEARCH Compensation and Benefits include:

• Competitive Salary
• Per Diem (for overnight travel)
• SEARCH-subsidized Group Health Insurance (including Medical, Dental, and Vision coverage)
• SEARCH-provided Short-Term Disability Insurance
• SEARCH-provided Long-Term Disability Insurance
• SEARCH-provided Group Term Life Insurance
• Voluntary Life Insurance
• Employee Assistance Program
• Emergency Travel Assistance
• Generous 401(k) Retirement Plan
• Paid Holidays (11)
• Paid Time Off (PTO)
• Merit-Based Raises, Bonuses, and Promotions