Research Staff, LLMs

WorkinVirtual-curated

Checked against the employer’s official careers page

WorkinVirtual wrote an original summary of this opportunity for remote-job discovery. We are not the employer and do not represent this company. Confirm the current requirements and complete the application only on the official employer page.

Official source last checked November 18, 2026.
Remote job overview

Remote — United States

Deepgram

Posted November 18, 2026 · Official source checked November 18, 2026

Limited information

Information labels describe listing completeness. Source-check dates and company-verification badges describe separate checks; they do not guarantee a vacancy, eligibility or hiring outcome.

Eligible locations
Not disclosed
Required time zone
Not disclosed
Employment type
Full-time
Experience
Not disclosed
Compensation
Not disclosed
Work authorization
Not disclosed
Travel
Not disclosed
Training
Not disclosed
Application
Company application site
Listing closes
December 7, 2026
Remote status declaredApply link checked

Deepgram is hiring a full-time member of its Research Staff to develop large language models for voice AI. The role focuses on research problems that arise when models must work across different voices, speaking styles and acoustic conditions, while remaining practical to train and deploy.

What you’ll do

  • Define new research projects with other researchers. Review and compare published methods to identify approaches worth testing for Deepgram’s products.
  • Design experiments that test the central hypothesis early. Analyze the results, decide which approaches to develop further, and plan how to scale successful prototypes.
  • Work on data curation, transformer architecture and reinforcement learning. Run large training jobs on distributed infrastructure, use the results to improve models and help deploy new models into production.
  • Document experiments and explain the results to the people who will use them. Follow developments in deep learning and evaluate their implications for voice products.
  • Build and evaluate your own AI-assisted research workflows. Deepgram asks interviewees to describe a workflow they built and explain how they reviewed its output.

What you’ll bring

  • At least three years in applied deep learning research, with experience comparing neural network architectures and loss functions and understanding their practical effects.
  • Hands-on LLM experience in data curation, large distributed training runs, transformer optimization and reinforcement learning.
  • Strong Python and PyTorch skills, experience with autoregressive or sequence-to-sequence models, and experience with distributed computing and processing data at scale.
  • A record of designing experiments and using their results to improve models, with the ability to communicate technical findings clearly.
  • Deep knowledge of causal language models, distributed training and inference, experience with reinforcement learning from human feedback, and published research on LLMs or deep neural networks are helpful.

Location and compensation

The role is remote within the United States. Deepgram publishes separate annual salary bands by level and location, with equity and a 10% annual bonus. For San Francisco, New York City and Seattle, the USD ranges are $169,000 to $212,000 for Research Scientist, $216,000 to $270,000 for Senior Research Scientist and $270,000 to $338,000 for Staff Research Scientist. Elsewhere in the U.S., the corresponding ranges are $149,000 to $187,000, $197,000 to $247,000 and $241,000 to $301,000.

Apply through Deepgram’s official job board. The employer says its recruiters contact applicants from deepgram.com email addresses.

Official employer application

Apply on the employer’s official site

This is a WorkinVirtual-curated discovery listing, not an employer-managed application. WorkinVirtual does not collect, copy, email, or forward your resume for this role. Opening the site does not mean you applied.

Sign in to save this job