Independent research · METAL Lab · 2025 to present

PEAK Self-Powered TinyML Sensor Node

A battery-free embedded sensing platform combining custom power hardware, firmware, on-device inference, and wireless communication.

My responsibility: sole developer of the hardware, PCB, embedded firmware, TinyML deployment, BLE reporting, experiment software, and manuscript preparation.

Back to selected work

Close-up of the independently designed PEAK research prototype — PEAK research prototype, independently designed and assembled.

172 nA: Measured power-board quiescent current
Sole developer: Hardware, embedded software, experiments, and paper
Custom PCB: Schematic, layout, assembly, and bring-up
Embedded TinyML: Firmware, inference, and BLE integration

Independent system development

I developed the complete research prototype across custom hardware, embedded software, on-device inference, and experiment tooling.

Designed the schematic and PCB, completed assembly and board bring-up, and iterated the hardware through measurement.
Implemented embedded acquisition, signal processing, task control, and TensorFlow Lite Micro inference.
Integrated BLE reporting and experiment software for repeatable system evaluation.

Research status

PEAK investigates energy-aware sensing and inference on battery-free embedded hardware. The prototype reaches 172 nA power-board quiescent current while integrating energy harvesting, embedded inference, and BLE reporting in one independently developed platform. The manuscript and full evaluation are in progress.