Experience

💼 Work Experience

• 2023-Current

  Principal Data Scientist, Capital One (Discover Financial Services)

  Discover Financial Services

  • Built Predictive ML models for borrower delinquency and early payoff risk using LightGBM and SHAP interpretability; enabled proactive servicing campaigns and reduced delinquency progression by 22%.
  • Developed an NLP model to automatically read and categorize borrower complaints, routing potential compliance issues to the correct servicing teams; reduced manual review time by 70%
  • Built a GenAI-powered Loss Mitigation Agent using RAG + LLM (LangChain) to summarize borrower cases, propose modification strategies, and draft compliant communications; cut analyst handling time by 45%.
  • Developed GenAI COPILOTS leveraging RAG, LangChain, and Snowflake to interpret regulatory controls and enable natural language–to–SQL querying for business users; automated exception monitoring and self-service analytics, reducing control-testing and ad-hoc analysis time by 50%.
  • Achieved a 95% reduction in fraudulent applications and sharply mitigated synthetic identity fraud risk by implementing a XGBoost anomaly detection model utilizing feature engineering on Adobe clickstream data.
• 2022

  Data Scientist

  Aspen Capital

  • Designed a strategy to offer second mortgages and credit cards to existing borrowers using TransUnion & CoreLogic datasets.
  • Modeled default probabilities and optimized lending rates using credit & behavioral attributes.
• 2021

  VP of Data Science

  PennyMac

  • Built end-to-end Loan Disposition LightGBM models (6 outcomes, 24-month horizon) → ≈$25M revenue in GNMA EBO space.
  • Developed Prepay/Default neural network on ~1 TB data to forecast conditional prepayment & delinquency probabilities.
  • Extensive EDA on macro & loan drivers (e.g., unemployment, hurricanes via Moody’s) showing nonlinear effects.
• 2020

  Investment Associate Intern

  CalPERS

  • Linear programming model to minimize index future roll cost.
  • Estimated incremental financing costs for stressed liquidity portfolios (collateral REPO vs OIS).
• 2019

  Quantitative Research Associate

  JP Morgan Chase & Co

  • LGD modeler for Wholesale Credit Loss Forecasting team.
  • Python log-linear model to estimate discount factors for Economic LGD prediction.
  • Sensitivity analyses to macro variables & recovery rates; impact studies in Pandas/SciPy/NumPy/Matplotlib.
  • Saved ~$9M capital via R-based NCO prediction model deployed on Linux.
  • Re-engineered LGD data processing to remove redundancy and improve efficiency.
• 2018

  Asset Management Intern

  Goldman Sachs

  • Advised a $2B AUM pension client in a multi-asset strategy team.
  • Optimized 20+ portfolios maximizing information ratio with leverage & vol selling.
  • Built an Excel tool to spot opportunities across 26 indices using earnings/sales data.
  • Benchmarked ETFs to define a dynamic cap-weighted REITs index.
• 2015–17

  Senior Associate (Python Developer)

  Bank of America Merrill Lynch

  • Python orchestrator on Quartz framework for derivative trade data processing.
  • Real-time settlement & reconciliation for global Equity & Forex derivatives.
  • Integrated money-market settlement tools across subsidiaries → process speed +150%.
  • Automated EOD reconciliation of nostro accounts → saved 300+ hours/month.

🎓 Education