At PayMe India, I built personal-loan underwriting models that cut new-customer NPA from 16% to 7% and existing-customer NPA from 7% to 4%. I engineered bureau features and mapped model scores to risk bands that drive approval cutoffs.
I also enabled automated underwriting, reducing approval time from days to 2–3 minutes. My work included an Early Warning System, lead-quality models, and a face liveness-detection model for digital onboarding.
At OLX Autos, I led data science for the Dealer Finance vertical, building limit-enhancement models and recommendation systems for asset-to-dealer assignment. At Home Credit India, I developed alternative-data underwriting models and propensity models, and designed and deployed OCR-based ML pipelines.
