Deepgram
Research Staff, Voice AI Foundations
$150K–$250K
USA - Remote +2 · Remote · Staff · Full-time · Posted Aug 23, 2026
- Location
- USA - Remote
San Francisco, CA
Ann Arbor, Michigan
About the job
Develop foundational voice AI research, including latent-space audio models, speech generation, scalable data methods, and efficient training and inference systems.
What you'll do
- Develop neural audio codecs, steerable speech-generation models, and embedding systems for controllable audio representations.
- Generate synthetic audio data and train multimodal speech-to-speech systems.
- Design model architectures, training schemes, and inference algorithms for efficient large-scale training and real-time deployment.
What they ask for
- Strong mathematical foundation in statistical learning theory, especially self-supervised and multimodal learning.
- Deep expertise in foundation-model architectures and scaling training across modalities.
- Ability to derive mathematical formulations and implement them efficiently; experience building quality-focused pipelines for massive datasets.
- Experience designing controlled experiments and optimizing models for real-world deployment, including hardware constraints and efficiency.
Nice to have
- Open-source contributions or research publications advancing speech/language AI.
Pay
$150K – $250K (base salary); 10% Annual Bonus (bonus amount)
AI focus
Research and develop latent-space voice AI, including audio codecs, speech generation, controllable representations, synthetic audio-data generation, multimodal speech-to-speech systems, and efficient large-scale training and inference. The company also expects employees to actively use and experiment with AI tools in their work.
Model-written summary · the original posting is authoritative