Predictive modeling and analytics work from my data science training.
Market Edge
AI-powered stock forecasting. LightGBM on next-day log returns, walk-forward cross-validation, full backtesting (Sharpe, max drawdown, equity curve vs buy-and-hold), and Claude-generated plain-English explanations of each forecast.
Python • Streamlit • LightGBM • Claude API • Plotly
Patient No-Show Prediction
Predictive model to identify patients likely to miss medical appointments. Includes EDA, class balancing, and feature importance analysis.
Python • Pandas • Scikit-learn
Churn Edge
AI-powered customer churn prediction. LightGBM with stratified k-fold cross-validation, engineered behavioral features, revenue-at-risk scoring, ROI-optimized retention campaign sizing, and Claude-generated retention recommendations per customer.
Python • Streamlit • LightGBM • Claude API • Plotly
Talent Pulse
Recruiting operations intelligence dashboard. Real-time visibility into hiring funnel, recruiter performance, and pipeline health: applicants, interviews, offers, hires, time-to-fill, and department-level deep dives with filterable date ranges.
Python • Streamlit • Pandas • Plotly
AI Therapist Chatbot
Empathetic mental health chatbot named Sage, powered by the Anthropic Claude API and LangChain. Uses thoughtful prompt design and conversation memory to support emotional reflection.
Python • LangChain • Streamlit • Anthropic Claude API