Summer Analyst
Date:
Worked as Data Analyst and Strategist in Cards Fraud Strategy Team under Marcus - Consumer divion of Goldman Sachs.
Built data-driven strategies backed with ML based models to capture fraud rings in credit card application space.
Objective
- Research working of fraud rings in credit card application space and build data-driven strategies to capture them
- Enhance fraud detection framework using ML techniques and articulate my insights into a business proposal
Strategy
- Devised 8 new parameters into framework to capture linkages between applications; uncover potential fraud rings
- Handled Big Tables with 15 million rows & applied ML models to design features, making framework more robust
- Built a univariate model based on Fraud Rate Analysis and Capture Rate Analysis on past data of PII linkages
- Devised a new mathematical Fraud Score to quantify riskiness of an application, to be used at pre-booking level
Impact
- Fraud Score saved additional 500K dollars per month for the company with higher Capture Rate & went into production
- Increased Fraud Hit Rate by 30% at pre-booking level & reduced False Positive by 25%, enhancing user experience
- Captured family fraud rings worth 84K dollars losses and fraud applications worth 50K dollars with huge income variation
My work in fraud detection was appreciated and converted into a full-time job offer at Goldman Sachs!