Senior/Staff Software Engineer - Frontier Research

Company Overview

Deepgram is bringing the world’s most natural interface, voice, to the technology we use every day. Our speech models and real-time infrastructure power AI systems that keep up with the way people actually speak. More than 5,000 organizations run conversational voice AI on Deepgram in production, including Twilio, Cloudflare, Sierra, Decagon, Vapi, Cresta, Granola, and Jack in the Box.

The Deepgram Voice Agent Platform brings together speech-to-text, text-to-speech and speech-to-speech, with voice-native foundation models available through cloud APIs or self-hosted and on-premises deployments. Built for accuracy, low latency and cost efficiency, Deepgram makes high-quality voice experiences practical to build and run at scale. More than 650,000 developers have used Deepgram to process more than 200,000 years of audio. Every day, more of the world’s conversations with AI run on our models.

Operating Principles:

Deepgram is a research company that productizes as fast as possible. Four Operating Principles shape how we think, decide, and work together:

1. Compress before you communicate. You think it through first, then lead with the bottom line.

2. Resist entropy. You question what already exists before adding more process or tools.

3. Solve the category. You fix recurring problems at the source, not one case at a time.

4. Work at the frontier. You go after problems no one has solved yet.

We look for people who already work this way, and these principles shape how we grow and evaluate our team.

AI-First by Default

We build AI that millions of people talk to, and we run the company the same way. AI is how we centralize work here, and is core to how we operate.

That means more than using the tools. We expect you to:

• Build, not just use. Design your own workflows and agents, and use AI to remove whole categories of work instead of doing the same work faster.

• Own the output. AI can speed up the work. It doesn’t make anyone less accountable for it. You understand, verify, and stand behind everything you share, including what an agent produces for you.

• Spend the time you get back at the frontier. The capacity AI creates goes toward problems no one has solved yet.

• Keep pace. Models and tools change every month, and so will your work. You adopt what’s new quickly and drop what no longer earns its place.

We judge AI by what it helps you deliver and whether you stand behind the result, not by how often you use it. In interviews, expect to walk us through a workflow you built and where you drew the line between AI and your own judgment.

Team Overview

Deepgram Research builds the speech-to-text and text-to-speech foundation models behind our platform. Today's scaling approaches won't deliver universal voice AI on their own. Audio is scarce, diverse, and high-dimensional, so we need new approaches. As the Research team's software engineer, you'll build the systems that make those experiments possible: training infrastructure, data pipelines, and evaluation tooling. You'll be an individual contributor working side by side with research scientists, data and infrastructure engineers, and product. Most of this work starts from a blank page, so you'll spend less time waiting on tickets and more time writing the spec.

Key Goals

• Shorten the STT/TTS model development cycle, from data prep through training and evaluation, by architecting horizontally scalable training systems

• Build high-throughput pipelines that ingest, version, and preprocess large, diverse sets of unstructured audio and text

• Own research infrastructure end to end, including distributed training, job orchestration, experiment tracking, and storage, in partnership with the Infrastructure team

• Ship automated evaluation tooling so researchers can validate or kill ideas quickly

• Turn promising research proofs-of-concept into production-grade systems that ship in Deepgram's STT and TTS models

• Build internal tools and dashboards that give non-technical stakeholders visibility into, and control over, ML workflows

• Set technical direction for research infrastructure by writing the design docs that define what gets built next

Minimum Skills, Knowledge & Capabilities

• Builds the first version of a system from a blank page with little direction, and writes the design or spec others build from

• Hands-on with large-scale distributed training, including multi-node GPU training and high-throughput data pipelines

• Fluent with container orchestration and ML lifecycle tooling such as Kubernetes, Docker, Prefect, and MLflow, or their equivalents

• Engineers data systems for large volumes of unstructured data, covering storage, versioning, and preprocessing at scale

• Has taken a research prototype from notebook to a system other people rely on

• Partners closely with researchers through shifting timelines and failed experiments, and can find the one experiment that settles a question

• Operates AI-native: uses and builds AI tools to amplify your own output, and can quickly build useful internal UIs with them

• Explains ML systems clearly to researchers and non-technical stakeholders alike

Preferred Qualifications (Nice-to-have, not required)

• Hands-on training or evaluating speech/audio models (ASR, STT, TTS), or LLM experience with a clear path to audio

• Has worked embedded with a research team where priorities shift with experimental results

• Contributions to open-source speech or ML tooling (e.g., ESPnet, NeMo, SpeechBrain, WhisperX)

• Typically ~5–12 years building software or ML systems, or equivalent demonstrated experience; advanced degree or equivalent industry experience.

We know the confidence gap and imposter syndrome can get in the way of meeting great candidates — so please don't hesitate to apply. We'd love to hear from you.

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 [email protected].