ASEAN GeoAI Fusion Hackathon 2026 (MCMC)
Beacon.ai (Disaster Escape)
Real-Time Disaster Preparedness & GeoAI Intelligence Platform
Real-time disaster preparedness and geospatial intelligence platform built for ASEAN GeoAI Fusion Hackathon 2026 (MCMC), combining NWS, USGS, and NASA FIRMS live feeds with Leaflet/OSRM turn-by-turn evacuation routing and a Python GeoAI sidecar for NDWI flood extent and SAM burn-scar segmentation.
System ID
03
Beacon.ai
Live Hazard APIs
M-013 Feeds
NWS, USGS, NASA FIRMS
GeoAI Sidecar
M-02NDWI + SAM
FastAPI + Python
Evacuation
M-03OSRM API
Turn-by-Turn Leaflet
AI Reasoning
M-04Gemini
Supabase Persistence
Section 01
Problem Statement
Civilians and responders during floods and wildfires need live hazard feeds coupled with satellite segmentation and turn-by-turn evacuation routes rather than static maps.
Section 02
System Constraints
- >Fuse heterogeneous live telemetry from NWS, USGS, and NASA FIRMS
- >Isolate heavy Python GeoAI inference (NDWI and SAM) from the web app
- >Provide interactive Leaflet turn-by-turn evacuation routing via OSRM
Section 03
System Architecture & Dataflow
NWS / USGS / NASA FIRMS
FastAPI GeoAI (SAM + NDWI)
OSRM Evacuation Routing
Next.js + Leaflet + Gemini
Section 04
Key Engineering Decisions
- 01Isolated NDWI flood extent mapping and SAM burn-scar segmentation inside a Python FastAPI GeoAI sidecar.
- 02Combined live NWS, USGS, and NASA FIRMS feeds with OSRM turn-by-turn evacuation routing in Leaflet.
- 03Integrated Gemini guidance with Supabase persistence in Next.js and Tailwind CSS.