At Leidos Inc., I modernize fraud models for the SSA Anti-Fraud program, porting legacy SAS models to Python and retiring five services and two pipelines. My AWS batch jobs keep more than 10 million records a day flowing to fraud models.
I built an AWS Athena pipeline that loads 21 million telecom records per month in 30 minutes. After a legacy data source was retired, I rebuilt a claims fraud model on payment data, reducing its inputs from 12 tables to three.
At GreenPortfolio, I improved its daily securities ETL, cutting cloud costs by 10% and processing time by 20%. I also deployed a forecasting pipeline with explainable outputs and prototyped embedding-based clustering for financial news.
I’ve also worked on client automation at HighRadius Corporation and clothing detection at Grroom. In projects, I’ve built a voice banking assistant, an in-memory key-value store, and a data-poisoning detection pipeline.
