Developed ML/deep learning models to automate document verification and catch suspicious IDs. Models use various methods to capture fraud including: shared template across cluster of IDs, PII information mismatch, signs of physical or digital manipulation.

Built transaction feature aggregator to characterize illicit transactions and compare users against public watchlists. These features were then used to train anti-money laundering and risk profiling models.

Leveraged data querying (SQL & Pyspark) and data visualization (Tableau & Domo) skills to guide stakeholders in making critical business decisions.