Rice Sentinel — Urban Sustainability Intelligence Agent

Inspiration

FIFA World Cup 2026 will flood 11 U.S. host cities with millions of visitors — hotels, venues, transit, and grids will spike at once. We wanted an early-warning agent that spots drift in Energy–Food–Water before a subsystem tips.

What it does

Rice Sentinel ingests Energy–Food–Water and air-quality time-series for FIFA 2026 host cities, scores each sensor against its seasonal baseline with a 3-model ensemble (strike Φ pUp=Φ(ln(spot/ref)/σ), momentum tilt, staleness detector), and surfaces the highest-impact intervention points. Generalized from predictrader-ai — same Φ k=√2/π, 120-sample ring.

Live: https://rice-sentinel.vercel.app · GitHub: https://github.com/binasalama12/rice-sentinel

How we built it

• Stack: Next.js 14, TypeScript, city-oracle.ts (drift ensemble), Vercel
• Methodology: UNLEASH Innovation Process — Problem framing → Ideation → Prototype → Test → Scale (see README UNLEASH table)
• Demo-safe: synthetic SYNTHETIC seed with dataSource flag — swap for live municipal feed

Challenges we ran into

Tuning z-score for non-trading data — city sensor baselines are seasonal, not spot price. Also translating trading FAILOVER to urban intervention ranking.

Accomplishments that we're proud of

• UNLEASH traceability: each phase mapped to repo artifact (README table)
• One base → 5 sentinels, all live 200 on Vercel — reuse >90%
• Pure cash $17.5k track, no credit padding — 327 participants

What we learned

Trading drift math transfers to city vitals — same ensemble, new domain. UNLEASH framing beats ad-hoc building for judge buy-in.

What's next for Rice Sentinel

Wire live municipal APIs (EFW + AQI), add district-level map, and pilot with one host city ops team. PRs open at binasalama12/rice-sentinel.

Built With

• next.js
• tailwind
• typescript
• vercel