Repository navigation
Overview
This release marks the original, unmodified source code submitted at the conclusion of LA Hacks 2026. Built entirely within a 36-hour sprint, this prototype serves as the Minimum Viable Product (MVP) for the Physio stroke rehabilitation wearable.
Demo 
Esp32.mp4
Devpost Submission: Click here :)
About LA Hacks
This project was built and deployed at LA Hacks, Southern California's premier annual student hackathon hosted at the University of California, Los Angeles (UCLA). The event brings together 1000+ engineers, designers, and creators from across the continent to prototype innovative hardware and software solutions under a 36 hour time constraint.
Key Features Built
- Real-time Sensor Data: Integrated dual MPU6050 IMUs utilizing AD0 hardware mapping to bypass I2C address collisions and capture precise joint movement.
- High-Frequency Telemetry: Engineered a pipeline streaming a custom 76-byte BLE payload at a stable 80Hz between the ESP32-C3 and the iOS frontend.
- Edge Machine Learning: Deployed an on-device CNN (90.5% accuracy) leveraging ZeticMLange and Apple vDSP to achieve a 0.04ms inference latency for real-time kinematic scoring.
- Hardware Prototype: Assembled a breadboard prototype with a custom-wired 3.3V regulated power rail to prevent sensor brownouts during BLE radio transmission.
Tech Stack
- Embedded: C++, ESP32-C3, I2C
- Frontend & ML: Swift, iOS, vDSP, ZeticMLange, Python
Contributors
- Scott Chian - Physical Hardware (Soldering), AI, Swift
- Ian Madden - Swift App Design & Creation + Zetic integration
- Ethan Pham - ML training & integration
- Majock Bim - Hardware Architecture & Embedded Systems
What's Next?
This codebase represents raw hackathon code. The next major release (v1.0.0) will include codebase refactoring + cleanup, and the integration of our custom sponsor-manufactured PCBA from PCBWay.