GridOptima
Turns an operator's plain-English notes into the cheapest valid 24-hour campus energy plan
My entry for the BUP CSE Fest 2026 hackathon's Smart Campus Energy Optimization Challenge. A campus runs on solar panels, a battery and the grid, and operators leave notes like "keep at least 40% in the battery overnight". GridOptima reads those notes with a language model, checks every rule it extracts with deterministic guardrails, then solves for the minimum-cost hour-by-hour schedule and replays it to prove it's valid before answering.