Harikesh Kushwaha account profile

Data Scientist with deep expertise in credit risk, fraud detection, and loan approval optimisation. Designed a CatBoost-based credit risk model using data from over 1 million customers, achieving a 25% reduction in NPAs and strong performance across key metrics (AUC: 78%, KS: 43%, Gini: 56%). Led the development and deployment of risk-based pricing systems and fraud scoring models, impacting millions of customers by improving ROI precision and reducing exposure to high-risk segments.

Focused on automating and optimising digital ad performance using Python and machine learning. Built end-to-end automation scripts for Amazon Marketing Services bidding, reducing manual effort by 50% and improving campaign performance by 20%. Implemented Airflow for scheduling, log tracking, and proactive issue detection, cutting issue identification time by 70%. Developed ML models to predict optimal bids, leading to a 30% reduction in cost-per-click (CPC) and improved ROI.