FloodLert AI
Innovation of Service
Flood risk, drawn on the map you’re looking at.
Predicts flood risk for the area on screen from live NOAA precipitation forecasts and local terrain. An ANUGA physics simulation does the real work; a PyTorch U-Net is the fallback when it can’t. Results land as colour-coded heatmaps.
- FastAPI
- React
- TypeScript
- MapLibre GL
- PyTorch
- ANUGA
Offline-first early warning for hospital wards.
One screen per nurse or physician showing which patient to see first, ranked by NEWS2 with each patient’s own baseline breaking ties. Built to work with no network: chest patch, ward gateway, edge server, cloud sync when it comes back. On simulated wards a distilled MindSpore model caught 56% of deteriorations in time against 36% for the NEWS2 chart trigger at a similar alert rate.
Simulator results, not clinical results.
- MindSpore
- Edge
- NEWS2
- Offline sync
A tool-calling agent that has to show its sources.
Bounded execution, corpus search, section retrieval and claim verification. Every response is grounded in source spans, and I built repeatable evals for retrieval recall@k and claim-support rate so “it feels better” never counts as evidence.
- LLM tool use
- RAG
- Evaluation
A Chrome extension that scores the page you’re on.
Rates phishing risk across 11 URL, content and scam-language signals, and explains itself on the page. Language classification with a heuristic fallback, plus local and shared caching so repeat checks don’t re-pay the inference latency.
- JavaScript
- FastAPI
- QStash
- Supabase
Lakad (LKD)
Grand Prize
Repo ↗
Smart urban mobility for walking Makati.
Hazard-aware walking routes, geotagged AR hazard reports, SOS video alerts and a city-side ticket dashboard. Built in 24 hours with the team that won the 2026 Poland–Philippines IT-Business Summit hackathon.
Race outcome prediction you can poke at.
Pick driver, circuit, tire compound and lap number; trained models return predictions on race outcome and performance. Mostly an excuse to make model behaviour legible to someone who isn’t me.
- scikit-learn
- HTML
- Dashboards
Sorting, spanning trees and recursion, with the numbers.
An analysis and simulation toolkit comparing sorting algorithms, visualizing MST construction and demonstrating recursion, with performance metrics next to every run.
Lending and inventory, kept boring on purpose.
A straightforward borrowing tracker. No models, no agents — just the CRUD app that the interesting projects quietly depend on knowing how to write.
Tracking intake without opening a spreadsheet.
A small Python tool for logging and watching protein intake over time. Built for one user, which is the correct number of users for most tools.
Nothing under that tag yet.