I build low-power Embedded ML systems and IoT firmware that are testable,
measurable, and built for real-world constraints.
>_

📍 Northern California

Currently: On-device voice control for smart lighting over Thread/Matter on Nordic hardware
Focus: Zephyr RTOS • Edge AI & Embedded ML • power profiling • maintainable firmware architecture

Local, privacy-first voice activation for Thread-based lighting control.

• On-device, Embedded ML-based keyword detection (no cloud dependencies)
• Thread/Matter-oriented system architecture for responsive local control
• Part of the harth modular AI system
• Repeatable power measurements and baselines

Repo: embr

Companion node for translating CoAP commands over Thread and coordinating Matter device behavior across a local mesh.

• Cornerstone of a modular, Edge AI-powered home intelligence system
• Clear interfaces and predictable behavior
• Built to be easy to extend as the system grows

• Layered firmware architecture (app / platform / HAL)
• Embedded TDD with mocks, stubs, and fakes for hardware behavior
• CI-first workflows (containerized builds + automated tests)
• Decisions and measurements documented (diagrams, ADRs, power data)

C/C++ • Zephyr RTOS + nRF Connect SDK • Edge Impulse • Nordic Semiconductor hardware • Thread • Matter

lux adyti, nexus machinae.
>_