Blaze-Guard

AI-driven wildfire detection, prediction, logistics routing, and citizen alerting platform built with a microservice architecture.

---

Overview

Blaze-Guard processes fire-related signals (satellite/event streams + ML outputs), routes decisions through specialized agents, and sends actionable alerts in near real-time.

Core goals:

• Detect potential wildfire incidents
• Predict spread and risk
• Dispatch fastest logistics routes
• Alert citizens and authorities
• Continuously improve model behavior via feedback loops

---

High-Level Architecture

• Frontend: React + TypeScript dashboard
• API Gateway: REST entrypoint for clients
• A2A Agent Layer:
  • Detection Agent
  • Prediction Agent
  • Logistics Agent
  • Citizen Alert Agent
  • Self-Evolving Agent
• Orchestrator:
  • agent registry
  • message routing
  • health monitoring
  • gRPC service
• Event Bus: Kafka
• Data:
  • PostgreSQL / PostGIS (geospatial logistics)
  • Redis / Redis Cloud (state/cache)
  • Oracle AI Database 26ai (AI agent memory / vector memory)

---

Repository Structure

Blaze-Guard/
├── README.md
└── Blazeguard/
    ├── docker-compose.yml
    ├── A2A/
    │   ├── go.mod
    │   ├── agent/
    │   │   ├── detection/
    │   │   ├── prediction/
    │   │   ├── logistics/
    │   │   ├── citizenalert/
    │   │   └── self/
    │   └── shared/
    ├── api-gateway/
    ├── Backend/
    │   └── Auth/
    ├── frontend/
    ├── kafka/
    ├── orchestrator/
    └── MD/

---

Services

1) A2A Agents (Blazeguard/A2A/agent)

• Detection: processes fire detection events
• Prediction: spread/risk estimation
• Logistics: nearest station query + Mapbox route optimization
• Citizen Alert: SMS/Push/Email-style alert flow
• Self Agent: quality/feedback monitoring for continuous improvement

2) API Gateway (Blazeguard/api-gateway)

• REST APIs for frontend/external clients
• CORS + middleware
• Publishes event payloads to Kafka topics

3) Orchestrator (Blazeguard/orchestrator)

• Agent registry/discovery
• Health monitor
• Routing layer
• gRPC service definitions and server

4) Auth Backend (Blazeguard/Backend/Auth)

• Authentication and user/session support
• Firebase + database integration

5) Frontend (Blazeguard/frontend)

• Authority + Citizen views
• Real-time dashboard + map workflows

---

Event Flow (E2E)

1. Data ingestion / ML generates detection or risk signals
2. Events published to Kafka (fire_detected, fire_prevention_check, etc.)
3. Detection/Prediction agents process and forward context
4. Logistics agent computes response route:
   • PostGIS nearest station filtering
   • Haversine prefilter (optional)
   • Mapbox shortest/fastest ETA route
5. Citizen Alert agent pushes emergency/prevention notifications
6. Self agent consumes outputs/feedback for model improvement signals
7. Frontend consumes APIs and visual updates

---

System Design

Prerequisites

• Go (1.22+ recommended; align with each go.mod)
• Node.js (18+ recommended)
• Docker Desktop
• Kafka (via docker compose in Blazeguard/kafka)
• PostgreSQL/PostGIS
• Redis (Redis Cloud supported)
• Mapbox API key

---

Environment Variables

Create .env files per service (do not commit secrets). Typical keys:

• KAFKA_BROKER=localhost:9092
• MAPBOX_API_KEY=...
• DB_HOST=...
• DB_PORT=5432
• DB_USER=...
• DB_PASSWORD=...
• DB_NAME=...
• DATABASE_URL=... (optional unified DB URL)
• REDIS_URL=rediss://...
• REDIS_PASSWORD=...
• ORACLE_AI_DB_DSN=...
• ORACLE_AI_DB_USER=...
• ORACLE_AI_DB_PASSWORD=...
• Firebase credentials/env for auth service

---

Local Setup

1) Start Kafka

cd d:\code-2-main\Blaze-Guard\Blazeguard\kafka
docker compose up -d

2) Start Orchestrator

cd d:\code-2-main\Blaze-Guard\Blazeguard\orchestrator
go mod tidy
go run .

3) Start A2A Agents

cd d:\code-2-main\Blaze-Guard\Blazeguard\A2A
go mod tidy
# Run each agent main.go in separate terminals

4) Start API Gateway

cd d:\code-2-main\Blaze-Guard\Blazeguard\api-gateway
go mod tidy
go run .\cmd\main.go

5) Start Frontend

cd d:\code-2-main\Blaze-Guard\Blazeguard\frontend
npm install
npm run dev

---

API Gateway Endpoints (Current)

• GET /health
• POST /api/v1/events/fire-detected
• POST /api/v1/events/fire-prevention-check

Payload example:

{
  "zone_id": "Z-101",
  "latitude": 28.6139,
  "longitude": 77.2090,
  "confidence": 0.92,
  "timestamp": "2026-03-25T10:20:30Z"
}

---

Key Kafka Topics (Current/Planned)

• fire_detected
• fire_prevention_check
• logistics_routes
• citizen_alerts
• self_agent_reports
• (optional) dead-letter topics for invalid events

---

gRPC (Orchestrator)

Proto location:

• Blazeguard/orchestrator/proto/orchestrator.proto

Main RPCs:

• Health
• ListAgents
• RouteMessage Use gRPC for internal service-control paths; keep REST for frontend compatibility.

---

Security Notes

• Secrets are ignored via .gitignore
• Firebase/GCP credentials must never be committed
• Use env injection or secret managers in deployment
• Prefer TLS endpoints for Redis Cloud (rediss://)

---

Current Status

• Multi-agent architecture implemented
• Kafka-first event flow available
• Orchestrator + gRPC scaffold present
• API Gateway base operational
• Deployment hardening and cloud wiring in progress

See:

• Blazeguard/MD/idea.md
• Blazeguard/MD/left.md
• Blazeguard/MD/logistics.md
• Blazeguard/problem.md

---

Contributors

• ML:Abhinandan
• Frontend: Rishank, Mahi
• Backend/Infra/Integration: Anshul

---

License

MIT (see LICENSE)