• Owned and executed CCAR (Credit Risk) models in PySpark which predicts customer default rates in collaboration with the ML team. Delivered default amounts of ~20M card holders/year to the finance team for annual risk assessment.• Served as the subject matter expert for regulatory model execution (CCAR, CECL, IFRS9), leading a team of three to ensure accurate and timely reporting during the 2022 fiscal year-end. Managed the technical execution of the models and guided the team through data… Show more • Owned and executed CCAR (Credit Risk) models in PySpark which predicts customer default rates in collaboration with the ML team. Delivered default amounts of ~20M card holders/year to the finance team for annual risk assessment.• Served as the subject matter expert for regulatory model execution (CCAR, CECL, IFRS9), leading a team of three to ensure accurate and timely reporting during the 2022 fiscal year-end. Managed the technical execution of the models and guided the team through data analysis and reporting.• Spearheaded a team in decommissioning SAS datasets and migrating to Python servers, resulting in significant space savings of approximately 9TB and achieving 10x faster record retrieval through optimized data handling.• Developed anti-money laundering (AML) reports for EMEA markets using SQL and rule-based models, analyzing flagged customers’ spending patterns and transaction history. Identified nearly 10% of reports as fraudulent and collaborated with strategy teams for further review Show less

• Migrated legacy regression models from SAS to PySpark, achieving a 30% reduction in execution time through the implementation of parallel processing on large datasets.• Created new variables leveraging SQL, including key performance indicators (credit and debit ratios), for ~20M records per month. Implemented alerts using statistical measures and established control limits to identify outliers and anomalies, reducing errors in the data pipeline by 90%.• Implemented an optimization… Show more • Migrated legacy regression models from SAS to PySpark, achieving a 30% reduction in execution time through the implementation of parallel processing on large datasets.• Created new variables leveraging SQL, including key performance indicators (credit and debit ratios), for ~20M records per month. Implemented alerts using statistical measures and established control limits to identify outliers and anomalies, reducing errors in the data pipeline by 90%.• Implemented an optimization initiative in Python that streamlined data processing tasks for regulatory models across 5+ markets and portfolios, including extraction, cleaning, feature engineering, and aggregation, resulting in a 50% reduction in execution time.• Developed a forecasting tool using PySpark and widgets that automates model executions, performance tracking and variance analysis. Thus reducing human intervention by 99%. Show less