AI / ML · case study
TrailMind
TrailMind is a voice and camera assistant that runs entirely offline on a Raspberry Pi 5. No cloud, no connectivity requirement: the models live on the device.
Why offline
Assistants stop working exactly where you might need them most: off-grid. TrailMind is built for that. Everything from speech to vision runs on-device, which also means nothing you say or show it ever leaves the hardware.
The on-device stack
A Raspberry Pi 5 handles the full loop, listening and responding by voice and understanding what the camera sees, inside the compute and memory budget of a single-board computer.
AI meets hardware
This is the intersection I care most about: taking models that normally assume a data center and making them work on constrained silicon. TrailMind applies the same instinct as my FPGA and embedded work to a self-contained product.