Embedded AI Engineer, On-Device Models
Deepgram · USA - Remote
Applied AIremotesenior / staff$219k–$274k
At a glance
Highlights
- Build the lowest layer of Deepgram's edge stack that makes new hardware viable.
- Bring world-leading speech models to embedded SoCs and purpose-built devices.
- Go below standard kernels and runtimes when they can't run models well enough.
What they offer
- Equity
- Bonus
- 10% annual bonus
Ideal candidate
A senior embedded engineer who lives close to the metal and wants to apply that craft to speech AI, with a builder mindset for ambiguous optimization problems and drive for measurable, production-quality results on constrained devices.
Requirements
- Hands-on experience with model optimization for on-device deployment, including quantization, pruning, knowledge distillation, or architecture-specific compilation.
- Familiarity with edge inference runtimes (e.g., ONNX Runtime, TensorRT, TFLite, ExecuTorch) and/or vendor-specific NPU/DSP toolchains.
- Strong understanding of hardware-software interaction — CPU/GPU/NPU/DSP architectures, memory hierarchies, fixed-point/integer arithmetic, and power management.
- Experience working close to the metal: bare-metal or RTOS environments, embedded Linux, or microcontroller and edge SoC development.
- Strong communication skills and a builder mindset to scope ambiguous optimization problems to measurable results.
- hiring process
- In interviews, expect to walk through a workflow you built and where you drew the line between AI and your own judgment.
Lab profile
About Deepgram
Real-time voice AI platform with speech models for conversational applications.
What they build
- Voice Agent Platform
- Speech models
- Real-time voice infrastructure
Known for
- Real-time speech processing
- Production-scale conversational voice AI
- Voice foundations and LLM research
Key facts
- 5,000+ organizations in production
Posted · Last seen