Agam Barta

AGAM BARTA tackles Bangladesh's fragmented multi-hazard crisis—disconnected Open-Meteo weather, USGS seismic, government advisories, AI predictions, and crowdsourced reports never unify to warn of floods, dengue outbreaks, road accidents, and network dead zones before they strike. I built a production-grade end-to-end platform: interactive multi-hazard map with 8 hazard types and filtering, address-based route risk analyzer with scoring, proactive geofenced push notifications, offline-first PWA with pre-downloaded hazard data for dead zones, one-tap emergency SOS with location sharing, crowdsourced reporting with live verification, and role-based moderation with audit logs. Millions are affected daily—rural farmers in flood-prone lowlands facing seasonal displacement, urban commuters on congested unsafe roads risking accidents, families in dengue-endemic neighborhoods vulnerable to outbreaks, elderly in remote offline villages cut off from alerts, and children in disaster-prone areas with no protection systems. It matters most during monsoon flash floods that submerge entire districts, peak dengue seasons when mosquito-borne infections spike, earthquake tremors that strike without warning, and daily commutes through hazardous routes—the critical window before entering danger zones when early warnings enable evacuation, alternative routing, mosquito source elimination, and life-saving preventive action before disaster arrives.

AGAM BARTA fuses 5 disparate data sources—Open-Meteo hyperlocal weather APIs delivering real-time precipitation, temperature, wind, and humidity data; USGS earthquake feeds detecting tremors above magnitude 3.0; official government disaster advisories from Bangladesh authorities; AI-driven predictive models using rule-based hazard forecasting from historical patterns; and crowdsourced citizen reports with photos, descriptions, and timestamps—into one unified geospatial intelligence engine with MongoDB geospatial indexing for fast location-based queries. The interactive multi-hazard map renders 8 distinct hazard types including floods, dengue hotspots, road accidents, earthquakes, extreme heat, thunderstorms, fog, and waterlogging with color-coded markers, interactive filtering by hazard type, distance display showing how far each hazard is from the user, and human-readable place names instead of raw coordinates. The address-based route risk analyzer uses OpenStreetMap Nominatim geocoding (proxied through backend) to convert addresses to coordinates, queries hazards within a configurable radius along the route corridor, and generates a composite risk score with breakdown by hazard category. Proactive geofenced push notifications leverage Socket.io real-time connections and Web Push API to send instant alerts when a user approaches a hazard zone, while the monitored locations feature lets users save important addresses and receive automatic alerts when new hazards emerge nearby. The offline-first Progressive Web App pre-downloads hazard data, cached map tiles, and static assets using service workers, ensuring full functionality in network dead zones where traditional alert systems completely fail. One-tap emergency SOS broadcasts the user's live location to pre-configured emergency contacts via SMS (Twilio), email (Nodemailer), and push notifications simultaneously. Community reporting empowers citizens to submit hazards with geolocation, photos, and severity ratings, with live verification workflows where moderators approve or reject reports. Role-based governance includes Admin, Moderator, and User roles with granular permissions, comprehensive audit logs tracking all actions, temporary ban system for offending users, and user management dashboards. Fully deployed on Vercel (frontend) and Render (backend) with MongoDB Atlas—live, functional, and production-ready today.

Languages: - JavaScript (ES6+) - HTML5 - CSS3 Frontend: - React 18 + Vite (UI Framework) - TailwindCSS 4 (Styling) - Leaflet + React-Leaflet (Interactive Maps) - Zustand (State Management) - Axios (HTTP Client) - React Router DOM (Routing) - i18next + react-i18next (Internationalization) - React Hook Form (Form Handling) - Socket.io Client (Real-time Updates) Backend: - Node.js + Express.js (REST API) - MongoDB + Mongoose (Database with Geospatial Queries) - Socket.io (WebSocket Server) - JWT + bcrypt (Authentication) - Node-cron (Scheduled Jobs) - Nodemailer (Email Alerts) - Twilio (SMS Fallback - Optional) APIs & Services: - OpenStreetMap Nominatim (Geocoding & Reverse Geocoding) - Open-Meteo API (Weather Data) - USGS API (Earthquake Data) - Web Push API (Push Notifications) - Service Worker API (Offline Support) Deployment: - Vercel (Frontend) - Render (Backend) - MongoDB Atlas (Database) - GitHub (Version Control)

AI Tools Used: - Claude (Anthropic) - Architecture planning and system design - ChatGPT (OpenAI) - Research, documentation, and query handling - DeepSeek - Debugging assistance and error resolution