Fraud Detection with Machine Learning: Developed a fraud detection model using LightGBM, achieving a Fraud Detection Rate (FDR) of 66.93% and projected annual fraud savings of $3.18 billion.

Comprehensive Data Cleaning & Feature Engineering: Processed a synthetic dataset of 1 million applications, addressing missing values, entity resolution, and behavioral patterns to enhance fraud identification.

Scalable & Adaptive Model: Implemented a periodic retraining strategy and extended detection capabilities to additional transaction types for improved fraud prevention in financial institutions.