kawacukennedy/community_voice_ews

Community Voice Early Warning System - Two-way disaster alert platform combining ICPAC official forecasts with crowdsourced SMS reports for the IGAD region (East Africa). Built for IGAD Hackathon 2026. # Community Voice Early Warning System A two-way early warning platform for the IGAD region — combining official ICPAC forecasts with crowdsourced SMS reports from local communities. No smartphone required. ```bash git clone github.com cd community_voice_ews make install && make dev-backend # Open localhost ``` Live Demo · API Docs · Report a Bug · Request Feature --- ## Why - **No connectivity required** — community members report via SMS (Africa's Talking / Twilio); internet not needed on their end - **Bilingual NLP** — automatically classifies reports in English and Swahili (flood, drought, pest, disease, fire, conflict, health) - **Official + Crowdsourced** — merges ICPAC satellite forecasts with on-the-ground reports - **Two-way communication** — authorities broadcast alerts back to all registered communities via SMS - **Free tier deploy** — costs nothing to run on Render free plan + SQLite; zero-config to start ## Features - **SMS Reporting** — community members text floods, droughts, pests, disease outbreaks — no smartphone required - **NLP Classification** — messages automatically categorized and prioritized in English and Swahili - **Live Map** — interactive Leaflet map with color-coded severity markers - **Two-way Alerts** — authorities broadcast SMS alerts to all communities in a region - **ICPAC Integration** — pulls official flood/drought/rainfall forecasts for cross-validation - **Offline-capable** — frontend works without internet after first load - **Dark mode** — automatically adapts to system preference - **Responsive** — works on phones, tablets, and desktops ## Quick Start ### Prerequisites - Python 3.12+ - pip ### Install & Run ```bash # Clone git clone github.com cd community_voice_ews # Backend make install make dev-backend ``` Open **http://localhost:8000** — the frontend is served automatically. API docs at **http:/ …