M.S. in Applied Data Science at the University of Chicago (Dec 2026), and B.S. in Statistics and Analytics from UNC-Chapel Hill. I work on risk analytics, machine learning, and LLM systems that turn messy operational data into decisions people can act on.
Selected work
- PPP Duplicate Loan Detection: entity resolution on 11M+ public Paycheck Protection Program loans to find businesses holding 2+ first-draw loans across lenders with no shared ID. DuckDB, LightGBM vs. Splink, and graph clustering.
- Virtual Engineer (UChicago Capstone): LLM-assisted alarm triage and fault investigation for campus building automation, built on BigQuery and Google Cloud.
- Customer Attrition Prediction: churn modeling on credit card customer data.
- MLOps Pipeline: MLflow experiment tracking and a Feast feature store.
- Abstractive News Summarization: Transformer (BART) summarization on CNN/DailyMail.
Tools: Python (pandas, scikit-learn, LightGBM, PyTorch), SQL, R, BigQuery / Google Cloud, Spark, DuckDB, Docker, Tableau, Power BI

