Applied ML Engineer - Edge Devices

The role

You will adapt Deepgram speech models to run on non-NVIDIA accelerators, edge servers, and embedded platforms. This involves modifying model structures, managing quantization, and building deployment pipelines for edge targets.

What you will do

• Port Deepgram speech models to non-NVIDIA and edge platforms
• Own serving-side model decisions like quantization and precision choices
• Validate every port on real hardware through benchmarks
• Build the deployment path for edge targets including packaging and conversion pipelines
• Work with Embedded AI Engineers to adapt models to custom kernels
• Partner with platform and silicon vendors on runtimes and toolchains
• Feed edge constraints back to Research and Impeller
• Take on adjacent production concerns like automated deployment and model security

Required qualifications

• Hands-on experience deploying ML models to edge or non-NVIDIA hardware in production
• Working knowledge of quantization and precision tradeoffs
• Experience with at least one edge or vendor inference runtime and its conversion toolchain
• Ability to modify a model to fit a platform by reading and rewriting model graphs
• Strong Python and PyTorch
• Production-quality engineering habits including tests, reproducibility, and benchmarks
• Comfort building automation around model conversion and deployment
• A builder mindset and clear communication

Nice to have

• Experience with speech, audio, or streaming/real-time models specifically
• Exposure to writing or reading low-level kernels (CUDA, Metal, NEON, or vendor DSP code)
• Experience with model security or integrity on deployed devices
• Familiarity with several accelerator families (Qualcomm, Apple, ARM, Intel, AMD, or custom NPUs)
• A track record of building internal tooling that made porting or deploying models faster

About the company

Deepgram is a research company that productizes speech models and real-time infrastructure for conversational voice AI. Their Voice Agent Platform provides speech-to-text, text-to-speech, and speech-to-speech via cloud APIs or self-hosted deployments.