Selected work

Projects.

Practical work across machine learning, data analysis, and software engineering—built around real questions and measurable outcomes.

SBA Loan Default Risk Predictor

An explainable XGBoost model for evaluating SBA 7(a) small-business loan default risk across roughly 8,000 Texas loans.

Achieved 94% accuracy, 0.964 ROC-AUC, and 81% recall on defaults.
Engineered 608 features from SBA FOIA, U.S. Census ACS, County Business Patterns, and Business Dynamics Statistics data.
Applied SHAP explainability and deployed the trained model with Streamlit.

PythonXGBoostSHAPStreamlitPandas

A responsive personal portfolio for presenting professional experience, projects, technical interests, and contact information.

Built reusable React components and responsive layouts with Next.js and TypeScript.
Created a centralized content model for experience, education, and project information.
Added swipeable photo cards, responsive navigation, and social sharing previews.

Next.jsReactTypeScriptCSS

Explore more of my code, experiments, and ongoing work.