Hassan Raza Mahmood

At i2c, I lead Data Science, Analytics, and Engineering for fraud and risk across 115 financial institutions. I own model strategy, production deployment, and model risk governance for authorization, enrollment, merchant, customer, and account-takeover risk.

I delivered seven production generations of an authorization fraud model that blocked $1.5M across 59M transactions in two months. The work raised per-client fraud capture from 8–10% to 34.6% of fraud count and 40.81% of fraud value while maintaining a 0.5% false-positive rate.

I design pooled XGBoost models, threshold economics, point-in-time feature pipelines, and real-time inference systems with SHAP reason codes. I’ve also used PyTorch LSTM embeddings and AI-assisted model review to improve detection and speed up challenger evaluation.

Before i2c, I built production malware and network anomaly detection pipelines at Broadstone Technologies. I lead a 15-person team, mentor people into independent project ownership, and work closely with Engineering, Product, Compliance, Governance, and clients to bring models safely into production.