TL;DR
This platform ingests heterogeneous IPDR logs, constructs communication graphs, and applies ML-based anomaly detection (94.16% accuracy) to deliver interactive visualizations. It features multi-format support, real-time processing, and end-to-end encryption.
Problem → Solution
Problem Manual analysis of massive IPDR data is inefficient and inaccessible to non-technical stakeholders, making it difficult to detect fraud and respond to security incidents quickly.
Solution An integrated platform that automates data parsing, applies ML for anomaly detection, and provides interactive graph visualizations for intuitive investigation and reporting.
Key Features
- Multi-Format Data Upload
- AI-Powered Anomaly Detection
- Interactive 2D & 3D Graphs
- Automated Relationship Mapping
- Comprehensive Reports History
- Search & Isolate Functionality
Architecture
A decoupled frontend and backend with a FastAPI serving ML models and a Next.js client for interactive visualizations.
Role & Credits
CIIS 2025 Hackathon Project This project was developed for the CIIS 2025 Hackathon to address the challenge of "Mapping A-Party to B-Party in IPDR Logs." I was part of a 5-member team, "Team Brigade," from VIT Bhopal University.
My Role: My primary responsibilities included developing the backend infrastructure with FastAPI, integrating the ML models, and designing the API for the frontend to consume.
Quickstart
# Clone the repository
git clone https://github.com/sujeetgund/ipdr-graph-engine.git
cd ipdr-graph-engine
# Backend setup
cd backend
pip install -r requirements.txt
uvicorn app.main:app --reload
# Frontend setup (in a new terminal)
cd frontend
npm install
npm start