Embedded AI Engineer: On-Device Speech & Edge Inference

Deepgram, Inc.; United States; On-site; USD 140,000 - 190,000; Full time

Job summary

Deepgram, Inc. is seeking an Embedded AI Engineer to bring speech models to on-device, low-power hardware. You’ll optimize for real-time inference across embedded CPUs/SoCs, write performance-critical code in C/C++/Rust, and integrate with edge runtimes.

You’ll collaborate with research and engine teams to push edge-friendly designs from day one and help ship private, offline voice experiences. We value hands-on experience with constrained hardware, knowledge of quantization and pruning, and a

Qualifications

• Experience delivering production systems on resource-constrained hardware.
• Strong proficiency in C, C++, and/or Rust for performance-critical code.
• Hands-on model optimization for on-device deployment (quantization, pruning, distillation).
• Familiarity with edge inference runtimes and NPU/DSP toolchains.
• Understanding of CPU/GPU/NPU architectures and power management for inference.
• Experience with bare-metal or RTOS environments and embedded Linux.
• Strong communication skills and a builder mindset to drive measurable results.

Responsibilities

• Take models to on-device, real-time inference across embedded processors and accelerators.
• Optimize for latency, memory, power, and thermal budgets via quantization and fusion.
• Write and optimize performance-critical runtime code in C/C++/Rust for embedded environments.
• Integrate with edge runtimes and vendor toolchains to map graphs onto hardware.
• Build on-device runtime plumbing: packaging, deployment, and OTA updates.
• Establish benchmarking and validation across target hardware to catch regressions.
• Collaborate with silicon vendors and research/engine teams to improve architectures.

Skills

C, C++, Rust

Edge AI on hardware

Model optimization

Edge inference runtimes

Hardware-software interaction

Bare-metal/RTOS

Communication mindset

Tools

ONNX Runtime

TensorRT

TFLite

FreeRTOS

Zephyr

Embedded Linux

Job description

Deepgram, Inc. is seeking an Embedded AI Engineer to bring speech models to on-device, low-power hardware. You’ll optimize for real-time inference across embedded CPUs/SoCs, write performance-critical code in C/C++/Rust, and integrate with edge runtimes.

You’ll collaborate with research and engine teams to push edge-friendly designs from day one and help ship private, offline voice experiences. We value hands-on experience with constrained hardware, knowledge of quantization and pruning, and a