Graph Neural Network AML Detection Engine

A production-ready Anti-Money Laundering (AML) detection system using Graph Neural Networks (GNNs) to analyze cryptocurrency transaction networks in real-time.

---

⚠️ Large Dataset & Backup Files

Some files required for this project (datasets and backups) are too large for GitHub and are hosted on Hugging Face.

Download them from: https://huggingface.co/IshaanPotle27/elliptic-aml-model

Place the downloaded files in the appropriate directories as described below.

---

Features

• Real-time fraud detection using GraphSAGE, GAT, and Temporal Graph Networks
• FastAPI backend and Streamlit dashboard for monitoring and explainability
• Scalable, production-ready Dockerized architecture
• Enterprise-grade security, monitoring, and compliance features

Quick Start

1. Clone the repository:

   git clone https://github.com/IshaanPotle/Graph-Neural-Network-AML-Detection-Engine.git
   cd Graph-Neural-Network-AML-Detection-Engine

2. Download large datasets and backups from Hugging Face (link) and place them in the correct folders (see above).
3. Run the Docker setup:

   ./scripts/docker-setup.sh

4. Access the services:
   • Dashboard: http://localhost:8501
   • API: http://localhost:8000
   • API Docs: http://localhost:8000/docs

Documentation

• Docker & Deployment: See README-Docker.md
• Improvements & Features: See README-Improvements.md

---

For more details, see the full documentation in this repository.