• Developed an AML model aggregating 500M+ asset movement entries over three years using Presto BDSQL and used Isolation Forest for anomaly detection across organization subgroups, leading to targeted recommendations for examiners to investigate firms exhibiting high-risk deviations.• Built TPS (Trading Person Searcher) - the first tool at FINRA enabling users to search across 60M+ customer accounts received throughout all departments, cutting investigation time by an estimated 3 weeks… Show more • Developed an AML model aggregating 500M+ asset movement entries over three years using Presto BDSQL and used Isolation Forest for anomaly detection across organization subgroups, leading to targeted recommendations for examiners to investigate firms exhibiting high-risk deviations.• Built TPS (Trading Person Searcher) - the first tool at FINRA enabling users to search across 60M+ customer accounts received throughout all departments, cutting investigation time by an estimated 3 weeks and enabling efficient querying across 200M+ records of customer data using PySpark with regex-based matching.• Designed and executed end-to-end ETL pipelines to integrate proprietary FINRA data of up to 7B+ records from Presto and DASL databases, utilizing distributed computing for efficient data processing, optimization, and automation of diverse pipeline workflows across varying time frames. Show less

• Used RStudio for feature reduction via L1 regularization, followed by an OLS fixed-effects model to identify 65 significant features for in-depth country-level analysis.• Conducted EDA and data cleaning on a dataset of 18,902 films with 2,180+ film element variables, incorporating nuanced 0-5 scaling and feature engineering techniques like min-max normalization and demeaning.

• Prospected and researched local dentist demographic and sent 2,500+ with 20% response rate offering financial services by using mail merge and excel to print labels of dentist addresses.• Contributed to the enhancement of financial models by reviewing and validating data accuracy, ensuring precision in their representation.