Smart-Farm Education Program

A hands-on education program commissioned by the Yangpyeong Education Office through the museum, where students experience an IoT smart farm and AI crop diagnosis first-hand. I planned the program and built everything it needed solo, a Python/Flask server, MicroPython firmware on the sensor devices, and the web app, designed so students watch data accumulate and sensor values change in real time, with their own eyes. The program is finished and was taught as a demo class to 5 high-school students in August 2026, and is scheduled to run in October 2026 as a class for a cohort of 15 Yangpyeong high-school students, selected by application.

A class runs as a guided flow: students enter the virtual lab, pick a room at the hub, and move through four STEPs (weather, AI vision, IoT, and camera), watching data load into the lab's Supabase database before finishing with 'my app complete' and a live hardware demo. A location-based weather API ties the lessons to real growing conditions, and students train a model themselves inside the browser, transfer learning with TensorFlow.js on a self-hosted MobileNet v2 feature extractor, so the epoch and loss numbers they watch are the real values from their own run.

The program is a museum × education-office commission: an exhibition-linked education case planned and built end to end by one person.