Find your next role
Discover roles across ambitious teams
20,349 jobs
- More jobs
Deepgram
Embedded AI Engineer, On-Device Models
Remote · USA · Senior level
$219k–$274k / year
About the role
Develop and optimize custom kernels and operators for embedded AI hardware platforms. Enable Deepgram's speech models to run efficiently on diverse edge devices with constrained resources.
What you’ll do
• Write and optimize custom kernels for non-NVIDIA accelerators and embedded SoCs. • Collapse models onto device execution units via quantization and operator fusion. • Integrate with vendor NPU/DSP toolchains and extend with custom operators. • Build performance-critical runtime code for embedded Linux, bare-metal, and RTOS. • Establish benchmarking and validation for latency, accuracy, and power.
What you’ll bring
• Experience with production systems on resource-constrained hardware. • Proficiency in C, C++, and/or Rust for performance-critical code. • Hands-on model optimization for on-device deployment. • Familiarity with edge inference runtimes and vendor toolchains. • Strong understanding of hardware-software interaction and embedded environments.
Life at Deepgram
Deepgram fosters an AI-first culture with rapid innovation and continuous learning. Employees actively use and build AI tools in their workflows.
About Deepgram
Deepgram provides real-time APIs for speech-to-text and voice AI, powering over 1,300 organizations with advanced voice-native foundation models.