Position: Principal Associate, Machine Learning

• Part of exclusive team for Discover Integration, reporting directly to leadership team. Currently working on low-volume testing involving the LNRV student/No Credit Model
• Upgrading/maintaining Kubeflow pipelines to optimize for 70M+ accounts for Capital One’s model (Thunder) for assessing credit risk associated with delinquency for authorizations for Upmarket and Mainstreet accounts
• Optimized spark configs to run ETL pipeline from 1 day to 6 months of data at a time for 3 years of retroscoring the credit risk model (Thunder) needed for deploying v2 to market
• Supporting Capital One’s in-market model (NPSL) for focusing on No Preset Spending Limit through Kubeflow & Spark optimizations
• Improved the efficiency of several Thunder model stages to go from erroring out to running saving hours for our DS partner
• Partnering with the data science team to resolve any issues they face when running experiments with the Kubeflow pipelines
• Operating as the first line of defense for pipeline failures for the credit risk models
• Co-lead of the Testing Education and Culture chair of the Code Excellence Initiative Committee to promote best practices. Working with Senior Managers to reviewing curriculum for pytest and property-based testing.
• CODA mentor for 3 in-coming Capital One employees, helping improve their experience here at Capital One
• Received award on triaging on unannounced breaking changes to saving significant downtown

Technologies: Databricks, SQL, New Relic, Python, Jenkins, Kubeflow, Databricks, Splunk, Spark, etc.