Designed for India's Gig Economy | Hackathon Project
GigShield (Full-Stack Event-Driven Platform) | Node.js, React.js, Express, MongoDB Live: guidewire-app.onrender.com
- Architected a MERN-stack platform with an automated trigger-evaluation engine; designed indexed MongoDB schemas across 5+ entity types, reducing trigger lookup latency as evaluation volume scaled.
- Engineered a real-time trigger engine monitoring disruption events and developed a dynamic premium calculator utilizing an AI-driven risk-scoring formula to ensure sustainable pricing.
- Built a high-fidelity, futuristic 3D interface to significantly enhance user trust and accessibility for non-technical gig partners.
To handle real-time data ingestion, seamless user interactions, and heavy machine learning processing, GigShield uses a modern microservices architecture:
The client application used by Delivery Partners and Admins.
- Built with React.js (Vite).
- Integrates React Three Fiber and Three.js for the futuristic 3D UI gateway.
- Contains the legacy HTML/JS dashboard under
public/for backward compatibility. - To run:
cd frontend && npm run dev
The core backend serving as the orchestrator for user data and API requests.
- Built with Node.js, Express, and MongoDB (Mongoose).
- Manages User Auth, Claims Storage, and routes ML requests.
- To run:
cd backend && node server.js
Dedicated service for Machine Learning and complex calculations.
- Built with Python, FastAPI, and scikit-learn.
- Contains the Isolation Forest ML model for fraud detection and risk scoring.
- Processes the real-time trigger engine (Weather APIs, AQI, Traffic).
- To run:
cd ml_service && pip install -r requirements.txt && uvicorn api:app --reload
Worker Registers on GigShield
↓
AI Risk Scoring (weather, AQI, zone risk, work pattern)
↓
Weekly Premium Calculated → WI × R × 0.05
↓
Trigger Engine polls APIs every 30 minutes
↓
Disruption Detected (rain / heat / AQI / flood / curfew)
↓
Multi-Signal Validation → Movement Check → Active Policy Check
↓
Fraud Detection Model (Isolation Forest) runs instantly
↓
If Approved → Payout credited to UPI instantly!
To run the full stack locally, you need to spin up all three services:
-
Start the ML Service (Port 8000)
cd ml_service pip install -r requirements.txt python main.py # Or run `uvicorn api:app --reload`
-
Start the Node.js Backend (Port 5000)
cd backend npm install node server.js -
Start the React Frontend (Port 5173)
cd frontend npm install npm run dev
Developed for DEVTrails 2026