Ashish Kesharwani

I’m a data scientist with 5 years of experience in credit risk modeling, lending analytics, and retail banking, translating signals into underwriting decisions.

At PayPal, I built XGBoost and decision-tree models for credit risk assessment and achieved a GINI coefficient of 0.58 and KS statistic of 40. I engineered 50+ underwriting features from bank-statement transaction data using Weight of Evidence (WOE), Information Value (IV), and segmentation, and I developed a decision-tree cutoff optimization decline strategy that identified high-risk segments with ~80% higher default rates. I also designed daily end-of-day balance features to flag 1.25% of funded accounts exhibiting 5.8 to 6 times higher delinquency rates over 30/60/90-day windows.

At WeRize (Wortgage Finance Pvt. Ltd.), I developed a personal credit score using regression techniques to predict delinquency risk and optimize loan amount and ROI. I delivered an 8.4% increase in average loan amount, a 9.4% reduction in 90+ days delinquency risk, and a 10% increase in conversion rate through targeted automation, continuous model validation with A/B testing, and clear stakeholder communication (including a dynamic Tableau dashboard).