Dallas-Fort Worth Metroplex
Sr. manager of Data Scientist with extensive hands-on experience in predictive modeling and statistical analysis. Strength includes, - Extensive experience in credit risk modeling, marketing analysis, model governance, statistical consulting and business analysis - Advanced knowledge and hands-on experience in big data analysis, linear/logistic regression, cluster analysis, decision tree, data mining, design of experiment, multivariate analysis
Leading Capital One Auto Finance (COAF) marketing and application stage decision modelings, including model design, development, implementation, documentation and monitoring. Leading Capital One Direct Auto Finance lost sale analysis, including customer behavior, dealer management, pricing test and credit policy Develop "who's in the market" model, help business optimize strategies across customer life cycle through maximizing customer relationship, understanding customer's need and providing timely offers. Develop marketing stage models, which using credit bureau information as well as credit card purchase history, for response and risk prediction in order to optimize marketing to acquire and retain the most valuable customers and avoid spending on poorly performing ones. Develop application stage risk models across different line of business, which using traditional three credit bureau information, deal structure characteristics as well as non-credit bureau information, for approve/decline decision make as well as pricing optimization. More than 10 billion dollars of auto loans are booked and priced using these risk models. Manage projects, assignments across different job families; Direct the outputs of direct report and interns.
Develop metrics and tools for automatic model monitoring. The tool handles models for marketing, application and loan servicing in Capital One Auto Finance. It produce monthly model monitoring report, including stability and validity monitoring for better model governance. Develop variable level monitoring tool for variable sanity check for data mining and model development Develop variable level impact tool to better understand the the variable level impact to final model score