Ruth AI Hackathon — 2nd of 150+ applicants
enerME
Raw energy data is almost unreadable to the person paying the bill. enerME ingests over two million rows of power readings with resumable pagination, then hands the user a forecast instead of a spreadsheet.
Prophet drives the forecasting layer, with spike detection classified into low, medium, and high severity. That severity rating is not just a color — it sets the tone of the AI-generated insight, so a minor blip reads calmly and a real anomaly reads urgent.
The design problem was not the model. It was deciding how alarmed the interface should sound.
Role
Full stack + data pipeline
Timeline
Hackathon weekend
Result
2nd place, 150+ applicants
Stack
React, TypeScript, FastAPI, Prophet, Supabase
Elsewhere