AI at the Edge

April 25, 2026 · 6:00 PM IST

TensorFlow Lite, ONNX, Raspberry Pi

Solo or team of 2

About This Theme

Run ML models directly on devices — no cloud, no latency, no privacy concerns. Edge AI is exploding: face detection on a Pi, wake word detection on a microcontroller, image classification on a phone. This hack night explores that frontier.

The Challenge

Deploy a machine learning model on an edge device (Raspberry Pi, phone, or browser via WebAssembly) that runs inference locally without calling an external API.

Push to a public GitHub repo · Submit before midnight · Solo or team of 2

Mentors for this Night

Jithin George

ML engineer, edge inference

jithingeorge

Mentors are available in the SHN Discord throughout the night.

Start Here

Not familiar with the stack? These resources will get you up to speed before Saturday.

TensorFlow Lite for Microcontrollers

docs

ONNX Runtime — Run models anywhere

docs

MediaPipe — On-device ML by Google

docs

Transformers.js — Run HuggingFace in browser

docs

Ready to build?

Free to join. No prior experience needed. Push to GitHub by midnight.