About this role
Deepgram needs a senior data scientist to tackle the complex infrastructure and engineering challenges presented by conversational audio, which requires analyzing rich speech data involving multiple speakers, accents, emotions, and acoustic conditions. Sitting at the intersection of research and data, this professional will build active-learning loops, design human-in-the-loop workflows, and create representative benchmarking datasets that directly drive model improvements. The ideal candidate brings an AI-first mindset, a high tolerance for ambiguity, and the ability to turn messy conversational data into systematic, scalable strategies.
Operating within a rapid, experimental environment, the person in this role will shape data operations from first principles and challenge conventional methodologies to uncover hidden signal. Success at Deepgram demands curiosity, a strong grasp of data pipelines, and a continuous desire to adapt while automating personal impact through advanced tooling.
What you will do
- Build comprehensive analyses to characterize internal data regarding languages, conditions, domains, speakers, and quality.
- Design and implement active-learning loops to systematically determine model priorities based on performance impact.
- Develop workflows, tooling, and model-assisted steps that optimize the utilization of human attention in the data loop.
- Create curated datasets and methodologies to enable representative benchmarking across diverse real-world speech conditions.
- Execute experiments that evaluate and challenge traditional data strategies to discover what truly drives performance.
- Maintain data consistency and cleanliness while ensuring information remains accessible to non-technical teams.
- Convert one-off, customer-specific model work into repeatable and documented pipelines for model adaptation.
What you need
- Bring hands-on experience working on real data pipelines and model-facing problems in data science, machine learning, or applied research.
- Demonstrate strong proficiency in Python and data tooling to independently build analysis, scoring, and automation systems.
- Possess background experience with data characterization, data selection, active learning, or related decision-making problems.
- Maintain working familiarity with speech, audio, or NLP models to reason about output quality, confidence, and error modes.
- Show a proven track record of translating ambiguous and messy data situations into measurable model or product improvements.
- Build reusable harnesses and systems that operate independently rather than relying on one-off notebooks.
- Communicate complex findings effectively to diverse audiences who may not share a technical background.
- Actively use advanced AI tools and be able to demonstrate personal applications and builds.
Nice to have
- Direct experience with ASR/TTS, audio data, or multilingual/code-switched data
- Experience with ensemble labeling, pseudo-labeling, or LLM-assisted annotation
- Familiarity with data provenance, PII/GDPR-aware pipelines, or model-improvement compliance
- Experience building custom or fine-tuned models for specific customers or domains
- Comfort working directly with research and engineering teams on shared infrastructure
Benefits
- Offers Equity
- Offers Bonus
Skills and tech stack
Questions candidates ask
Where is this role located?
The position is based in the USA with a remote work arrangement. Candidates can work from home while collaborating with distributed teams.
What is the compensation range for this position?
The salary range is between $165K and $220K. Additional compensation includes offers for equity and bonuses across multiple ranges.
What kind of operating rhythm does the company expect?
Deepgram expects an AI-first mindset where active use and experimentation with advanced AI tools are core to everyday work and performance measurement. The company moves at a rapid pace of change, requiring continuous learning and adaptation rather than a traditional 9-to-5 routine.
Who does this position report to?
This is a hands-on, high-leverage role that reports directly to the VP of Data Operations. The position offers unusual latitude to define an area from first principles.