Aegis Disaster Platform
An emergency preparedness and response platform combining real-time global satellite hazard feeds, verified NDMA/FEMA protocol guidance, and multi-channel disaster alert coordination.
- Role
- Lead Full Stack Architect (SIH 2025 Finalist)
- When
- 2025 – 2026 · Top 10 National Finalist
- Built with
- Node.jsExpress.jsMongoDB AtlasGemini 2.5 FlashGDACS Satellite APIViteTailwind CSSLeaflet.jsNodemailerGSAP
- Links
Aegis Disaster Platform
An emergency preparedness and response platform combining real-time global satellite hazard feeds, verified NDMA/FEMA protocol guidance, and multi-channel disaster alert coordination.
Why it exists
During major natural disasters like earthquakes, flash floods, or cyclones, standard municipal portals and messaging channels experience critical bottlenecks. First responders and citizens struggle to separate unverified rumors from official safety protocols.
Most emergency response applications are either static text archives or bloated portals that require heavy bandwidth and complex user onboarding. In emergency situations, every second spent navigating convoluted interfaces or deciphering unverified advice poses direct safety risks.
Aegis was conceived and built for Smart India Hackathon 2025 to solve this exact communication breakdown: providing a unified, lightweight command center that synchronizes live satellite telemetry from GDACS with official government disaster management protocols.
What it does
Aegis unifies global satellite hazard ingestion, verified AI disaster advice, and multi-channel emergency broadcast dispatching into an accessible web platform.
- GDACS Satellite Telemetry Ingestion: Continuously polls and parses GeoJSON hazard telemetry from the Global Disaster Alert and Coordination System (GDACS), pinpointing real-time coordinates, severity levels, and impact radiuses on an interactive Leaflet map.
- Protocol-Grounded RAG Assistant: Integrates Google Gemini 2.5 Flash grounded strictly in official National Disaster Management Authority (NDMA) and FEMA emergency action manuals, answering emergency questions with zero medical hallucinations.
- 8 Specialized Disaster Safety Modules: Dedicated interactive safety workflows for Earthquakes, Floods, Cyclones, Landslides, Tsunamis, Droughts, Heatwaves, and Fires with bilingual guidance and printable offline action checklists.
- Campus Emergency Broadcast Dispatcher: Institutional command interface allowing incident commanders to dispatch urgent safety bulletins, evacuation coordinates, and alert notices directly to registered students via Nodemailer.
- Fault-Tolerant Student Access: Built with multi-factor OTP fallbacks and resilient session caching on MongoDB Atlas so users in disrupted network environments never face login lockouts during active emergencies.
What it won’t do
Aegis enforces strict technical and ethical boundaries designed around disaster response safety standards.
- No Autonomous Medical Prescriptions: Aegis strictly refuses to recommend prescription medications, triage dosages, or chemical treatments. In life-threatening medical situations, it directs users exclusively to official emergency numbers and basic first-aid stabilization checklists.
- No Unverified Rumor Ingestion: Hazard alerts are strictly accepted only from certified government and meteorological APIs (GDACS, IMD). Social media scraping or unverified crowd reports are rejected to eliminate panic-inducing false alarms.
- Zero Plaintext Telemetry or Credential Storage: User credentials and campus contact telemetry are strictly hashed using bcrypt and isolated within secure MongoDB Atlas clusters with role-based access control.
How it works
The architecture is organized as a decoupled full-stack platform with a performant Express REST backend and a responsive frontend client.
Geospatial Telemetry Processing: The server regularly polls GDACS GeoJSON feeds, filtering hazards by geographic bounding boxes and severity thresholds. Events are normalized into lightweight coordinates sent to client map layers without overloading mobile client bandwidth.
Grounding Constraints: When a user queries the emergency AI assistant, the query is combined with official NDMA/FEMA procedural context in an isolated inference envelope. If an inquiry falls outside emergency preparedness (e.g. asking for specific pharmaceutical dosages), the model is instructed to output an explicit refusal and redirect to emergency helpline contacts.
Resilient Offline Strategy: Disaster modules and emergency contact cards are cached using service worker storage, ensuring that users with spotty cell reception can still reference life-saving checklists.
By the numbers
What’s next
- Offline SMS Gateway: Integrating Twilio/GSM hardware fallbacks to deliver alert dispatches to basic non-smartphone devices without active cellular internet.
- Community Crowdsourced Hazard Verification: Introducing peer-reviewed check-ins with multi-signature verification from nearby local authorities before surfacing hazard pins.
Enterprise lead conversion scoring and predictive analytics engine