Focused on optimizing Deepgram's speech models for low-power consumer hardware, the full-time Embedded AI Engineer will enhance on-device inference across various processors and accelerators while working remotely.
Key responsibilities
Optimize and compile speech models for embedded devices, ensuring real-time performance and efficiency
Write and refine performance-critical runtime code in C, C++, and/or Rust for constrained environments
Establish benchmarking and validation processes to measure model performance across target hardware
Required qualifications
Experience with production systems on resource-constrained hardware, such as embedded systems or mobile devices
Strong proficiency in C, C++, and/or Rust, specifically for performance-critical applications
Hands-on experience with model optimization techniques for on-device deployment
Familiarity with edge inference runtimes and vendor-specific NPU/DSP toolchains
Understanding of hardware-software interaction and its impact on inference performance