upi-security

Autonomous Red vs. Blue AI Fraud Simulation & Defense System for UPI & Digital Payments. Features LLM-driven zero-day attack generation (F3 Taxonomy), PyTorch Bidirectional LSTM + LightGBM + GraphSAGE multi-model defense, and real-time SIEM dashboard.

Contextual payment security engine for UPI transfers — mitigates coercion and panic fraud using personal baseline metrics (Median/MAD), 6 calibrated risk signals & 4-tier adaptive friction.

Static-first UPI scam-awareness prototype with a browser demo, Python reference engine, and Kotlin/Compose Android prototype; it does not intercept live bank transactions.

Real-time Mule Account Detection Engine for UPI. Uses Graph Neural Networks (GraphSAGE), JA3 Fingerprinting, and Behavioral Forensics (EIF) to identify money laundering rings, smurfing, and collusive fraud networks in milliseconds.