In Progress

Small Business Due Diligence AI

An AI-powered platform that helps first-time buyers evaluate small businesses (under $5M). Upload messy seller documents — get structured red flags, normalized cash flow analysis, and a professional deal memo. Built with Next.js, Claude API, Supabase, and Stripe.

Goal: 100 users · $15–25K MRR · Broker partnerships
MSBA 204 · Python, R, SQL

LP Portfolio Optimizer

Linear-programming model that maximizes expected return subject to risk, diversification, and budget constraints. Auto-fetches ETF price data from Yahoo Finance, solves with SciPy (Python) and lpSolve (R), stores results in SQLite, and compares four risk scenarios in a Streamlit dashboard. The base case allocates six ETFs for a 1.40% expected monthly return at 3.00% risk.

MSBA 207 · Machine Learning

Loan Acceptance & Default Risk Models

Logistic regression and binomial GLM models in Python (pandas, scikit-learn, statsmodels) to predict consumer-loan acceptance and SBA loan default. Compared model configurations, interpreted coefficients and AIC, and evaluated classifications with confusion tables and misclassification rates.

Live

This Website

Built from scratch with vanilla HTML, CSS, and JavaScript. Deployed on Cloudflare Pages via thanh.lyforge.net. Simple, fast, and maintainable — because sometimes the best tools are the simplest ones.