Projects

AML Detection with XGBoost & SHAP

A machine learning pipeline for detecting money laundering in financial transaction data, built on the IBM Synthetic AML dataset (5M+ transactions). Combines a time-based train/val/test split to prevent temporal leakage, an XGBoost classifier, and SHAP for transaction-level explainability — built with the Swiss regulatory framework (revDSG, AMLA) in mind.