🛡️ Nigehban AI

Autonomous National Crisis Intelligence & Resource Mobilization Platform

Note

Nigehban AI (formerly AegisNet) was built for the Google AI Hackathon. It represents a state-of-the-art emergency operations paradigm that coordinates autonomous multi-spectrum intelligence feeds (Meteo, GDELT, social streams) to intercept, classify, and mitigate national crises across Pakistan.

---

🌍 Overall System Design & Architecture

Nigehban AI operates as a decoupled, real-time, event-driven ecosystem composed of a Spring Boot Reasoning Engine, a keyless Multi-Spectral Ingestion Swarm, and real-time Tactical HUDS (Angular Web + Kotlin Mobile App):

graph TB
    subgraph Ingestion Swarm [Multi-Source Data Ingestion]
        OM[Open-Meteo API] -->|Meteo / Air Quality| AMS[Autonomous Monitor Service]
        GD[GDELT Project API] -->|Global Crisis Media| AMS
        GC[GDACS EU Feed] -->|UN Disaster Telemetry| AMS
        FM[NASA FIRMS] -->|Thermal Anomalies / Fires| AMS
        PM[PMD CAP Alert] -->|Local CAP Warnings| AMS
        HX[UN HDX Portal] -->|Hydrological Droughts| AMS
        BS[Bluesky RPC] -->|Live Social Keywords| AMS
        MA[Mastodon API] -->|Hashtag Timelines| AMS
    end

    subgraph Core Brain [Spring Boot Core Engine]
        AMS -->|Structured Threat Stream| CIA[Crisis Intelligence Agent]
        CIA -->|Threat Mapping Taxonomy| CIA
        CIA -->|Compute Criticality & Dispatch Plan| CIA
        ST[Simulation API] -->|Mock Social/Weather Payloads| CIA
        REST[Direct User Uplink] -->|Citizen Incident Report| CIA
    end

    subgraph Broadcast Layer [STOMP over WebSockets]
        CIA -->|/topic/crisis-events| WebHUD[🗺️ Web Tactical HUD]
        CIA -->|/topic/city-threats| WebHUD
        CIA -->|/topic/traces| WebHUD
    end

    subgraph Client Interface [Synced Operations Ecosystem]
        WebHUD -->|Geospatial Satellite Maps| HU[1. Tactical Map HUD]
        WebHUD -->|Resource Dashboards| RC[2. Resource Center]
        WebHUD -->|Citizen Uplink Forms| CR[3. Citizen Reporting Portal]
        Mobile[AegisNet Kotlin Mobile Client] -->|On-Device Gemini Ingest| REST
    end

    classDef real fill:#1c2833,stroke:#3b82f6,stroke-width:2px,color:#fff;
    classDef core fill:#0f172a,stroke:#f43f5e,stroke-width:2px,color:#fff;
    classDef client fill:#0f172a,stroke:#10b981,stroke-width:2px,color:#fff;
    class OM,GD,GC,FM,PM,HX,BS,MA real;
    class AMS,CIA,ST,REST core;
    class HU,RC,CR,Mobile client;

Loading

For a comprehensive architectural breakdown and code-level walkthroughs, please refer to the main documentation: 👉 Nigehban AI Engineering Documentation

---

🧠 The Multi-Agent Orchestration Swarm

The platform aggregates noisy global data by orchestrating six distinct autonomous agents running on dedicated threads:

1. Agent 1 (Meteorological & Environmental): Polls Open-Meteo every 30s. Monitors temperatures, flash floods (river flow discharges), cyclones, and PM2.5 air quality indexes.
2. Agent 3 (News Intelligence): Polls the GDELT Project every 90s, translating raw media reports into classified geopolitical/infrastructural threats.
3. Agent 4 (Decentralized Social Signals): Scrapes Bluesky and Mastodon public timelines every 60s to identify rapid localized incident spikes.
4. Agent 5 (Advanced Technical Intelligence): Cross-references NASA FIRMS fire models, PMD CAP advisories, and UN/OCHA HDX hydrological data every 120s.
5. Agent 6 (Global Tectonic Disasters): Pulls GDACS feeds every 5m to verify high-magnitude seismic epicenters.
6. Central EOC Reasoning Agent: Computes dynamic criticality metrics, estimates affected populations, maps emergencies to the Pakistan National Hazard Taxonomy, and structures automated ground dispatch plans.

---

📡 Keyless API Integration Matrix

The ingestion layer runs on keyless, unauthenticated real-time channels combined with automatic, high-fidelity mock fallbacks to guarantee absolute uptime:

• Real APIs Used:
  • Open-Meteo (api.open-meteo.com)
  • GDACS EU Joint Research Centre Feed (gdacs.org/xml/rss.xml)
  • Bluesky Social search (public.api.bsky.app/xrpc/app.bsky.feed.searchPosts)
  • Mastodon tag timelines (mastodon.social/api/v1/timelines/tag/:tag)
  • NASA FIRMS, PMD CAP, and UN/OCHA HDX open feeds.
• Mock & Fallback APIs:
  • GDELT Fallback: GDELT frequently rate-limits (HTTP 429). GdeltService.java automatically catches connection errors and injects a Pakistan-specific taxonomic news database (e.g. Gilgit Glacial Lake Outburst Floods, Sukkur Dam failures) to guarantee system execution.
  • Simulation Injections: /api/simulation/weather and /api/simulation/social allow instant manual demo triggers representing extreme weather alerts and social panic posts.
  • Citizen Uplink REST Portal: /api/simulation/report receives direct JSON coordinates from the Angular Web Reporting tab or Android Kotlin client, immediately generating pulsing warning rings on the map.

---

🛠️ Technology Stack

• Reasoning Engine: Java 21, Spring Boot, Spring WebSockets (STOMP), SockJS, Maven, Lombok
• Tactical HUD (Web): Angular 17, TailwindCSS (Dark Mode), Leaflet Spatial Engine, RxJS, SockJS-Client, StompJS
• Mobile Interface: Android, Kotlin, Jetpack Compose, On-Device Gemini Nano Integration

---

🚀 How to Run Locally

1. Launch Backend Engine

cd backend
# Set Java 21 runtime
export JAVA_HOME="/path/to/your/jdk-21"
mvn spring-boot:run

The Tomcat container boots on port 8080 and launches the monitor polling cycles.

2. Launch Tactical Web HUD

cd frontend
npm install
npm start

The dashboard will compile and open at http://localhost:4200 with real-time websocket synchronization.

3. Launch Mobile Companion Client

1. Open the /mobile directory in Android Studio.
2. Build and run the project in an emulator or active hardware.