Data scientist building production ML systems at the intersection of finance and machine learning — credit risk, survival analysis, anomaly detection, and forecasting. I focus on the full stack: from model architecture through API serving and interactive front-ends.
Currently building
- LoanSurv — Survival analysis for loan default timing. Cox PH + Random Survival Forest on 2.2M Lending Club loans. Runs entirely in the browser via in-browser inference. [live demo]
- Credit Recourse Engine — Most credit models tell you who defaults. This one tells denied applicants what to change. XGBoost + MAPIE conformal prediction + DiCE counterfactual explanations.
Stack
ML/Data Python · scikit-learn · XGBoost · lifelines · scikit-survival · PyTorch
Data pandas · SQL · Spark · Azure Synapse · Cosmos DB · pyarrow
APIs FastAPI · Pydantic · Docker · Azure Stream Analytics
Frontend React · Vite · Recharts · Tailwind CSS
Other projects
| Project | What it does |
|---|---|
| opsmon | Operational monitoring framework — anomaly detection, drift alerts, reliability cards |
| scenario-planning-analytics | Urban demand forecasting for city service planning — Prophet + XGBoost |
| controlled-document-retrieval | Hybrid compliance document retrieval with jurisdiction filters and audit citations |
| azure-stream-analytics | Real-time event stream processing with Azure Event Hubs and Power BI |
| MBS Prepayment | Predicting mortgage-backed securities prepayment risk |