A comprehensive fraud detection system built with FastAPI backend and Next.js frontend, utilizing Aerospike Graph for real-time graph-based fraud detection.
New to this project? Check out our detailed Setup Instructions for complete installation and configuration guidance.
Already set up? Run the application:
./run_app.shAccess the application:
- Frontend: http://localhost:4001
- Backend API: http://localhost:4000
- API Documentation: http://localhost:4000/docs
- Framework: FastAPI with async support
- Graph Database: Aerospike Graph Service (AGS) via Gremlin queries
- Features:
- RESTful API endpoints for fraud detection
- Real-time Gremlin query execution
- Sample data seeding
- User and transaction management
- Fraud pattern detection algorithms
- Framework: Next.js 14 with App Router
- Styling: TailwindCSS with dark/light theme support
- Features:
- Modern, responsive dashboard
- Real-time data visualization
- User and transaction exploration
- Fraud pattern analysis
- Interactive graph visualization (Phase 2)
The system implements real-time fraud detection using graph-based analysis:
- Purpose: Detects transactions involving previously flagged accounts
- Method: 1-hop graph lookup for immediate threat detection
- Risk Level: High
- Use Cases: Known fraudster connections, blacklisted accounts
- Purpose: Detects transactions involving accounts connected to flagged devices
- Method: Network analysis through transaction history
- Risk Level: High
- Use Cases: Device-based fraud networks, shared device abuse
- Purpose: Identifies accounts with unusually high connectivity
- Method: Graph centrality analysis
- Risk Level: Medium-High
- Use Cases: Money laundering hubs, distribution networks
- Setup Instructions - Complete installation and configuration guide
- Data Model - Detailed data structure documentation
- RT1 Fraud Detection - RT1 implementation details
- Project Plan - Development roadmap and milestones
This project is licensed under the MIT License - see the LICENSE file for details.