"ResQ team on IBM Bob 2.0 hackathon Hackathon"
Team Idea
Our system is a landslide early-warning platform for India's North-Eastern Region, built for SIH 2026 (PS26001). It combines two layers: a machine-learning susceptibility model — a Random Forest trained on real Geological Survey of India landslide records, validated with spatial cross-validation (ROC-AUC 0.735) — and a rule-based rainfall alert engine that fires automatic SMS and voice-call alerts via Twilio when live rainfall crosses a published threshold for a high-risk zone. Citizens can report warning signs (cracks, water seepage, tilting trees) through an offline-capable mobile form in English, Hindi, or Nepali, with AI-assisted translation and summarization for officers. Officials get a live, multilingual dashboard showing zone risk, active alerts, and citizen reports, with a confirmation step before any serious broadcast. The architecture is built to scale to all eight NER states; Sikkim is currently the only state with real, populated data and a trained model.
Team Members
.png&w=256&q=75)
Rushitha Borra
Riaa
.png&w=256&q=75)
Vugiri Govardhan
Govardhan_1757
Submission
Team Leader hasn't made a submission yet