BioCycle Djerba
🌱 Turning hotel food waste into renewable energy. Djerba's hotels dump
tonnes of organic waste every week; captured in a biomethanization plant, that same methane is
electricity. The gap was never the science, it was logistics: which pickup is worth sending a truck for,
today? So we built the platform that answers it.
FastAPI
PostgreSQL
React
TypeScript
TailwindCSS
Djerba's hotels generate tonnes of organic waste every week. Most of it goes to landfill, where it
releases methane straight into the atmosphere. That same methane, captured in a biomethanization
plant, is electricity. The gap was never the science. It was logistics: which pickup is worth sending
a truck for, today? So we built the platform that answers it.
How it works:
🏨 Hotels declare their organic waste in a few taps
🤖 An AI layer scores each load: quality, purity, contamination, expected methane and energy
yield
📋 Operators get a ranked collection queue instead of a first-come-first-served inbox
🗺 Every request opens onto a map showing the hotel, the plant, and the distance between
them
♻ Each pickup is tracked end to end: pending → accepted → on the way → collected
→ completed
The part I'd underline:
We deliberately did not let the AI's opaque priority
score drive the running order. Operators dispatch real trucks, and "because the model said so" is not
a reason anyone can act on. The queue sorts on rules a human can read and challenge: waste quality,
distance to the plant, load size, then first-in-first-out. The AI informs the decision — the
operator makes it.
Under the hood:
FastAPI · PostgreSQL · SQLAlchemy 2.0 (async)
· React · TypeScript · TailwindCSS · Leaflet · Docker