At American Express, I build credit and fraud risk models that improve prospect prioritization and acquisition targeting. My XGBoost prospect spend model delivered an estimated $3.9M GCM impact, improving prediction accuracy from 80% to 96%.
I engineer and evaluate large feature sets, apply SHAP for model interpretability, and validate models across in-sample, out-of-sample, and out-of-time datasets. I’ve also developed revolving-behavior models that improved prediction accuracy by 23% and built Custom GPTs to make model knowledge more accessible across the CFR team.
Previously at Axis Bank, I automated reporting workflows with PowerApps, Power Automate, and Oracle SQL, reducing turnaround time by 80%. I also bring product-analysis experience from Nearbuy, where I translated usability and competitive insights into recommendations that improved product quality.
