An autonomous LLM-powered assistant that reasons about your environment (calendar, biometrics, devices, and habits) to execute multi-step routines without being explicitly asked. Orchestrates 10+ real integrations (Spotify, Google Calendar, Garmin, smart home devices) via a tool-calling agent loop. Trains local stress and sleep predictors on wearable data using scikit-learn, with a confidence system that learns from feedback and adapts over time. Built with a FastAPI backend, React frontend, and real-time WebSocket layer.

A medical triage system that uses LLM semantic parsing to convert free-text symptom descriptions into structured clinical features, then applies an ML classifier to produce risk scores and triage decisions. Achieves 0.92 F1-score on patient risk classification. Built with a FastAPI backend, SQLAlchemy-managed SQLite store, and a live Streamlit dashboard.

A causal inference application using Double Machine Learning (DoubleML) to isolate the true return on marketing spend by controlling for confounding variables to separate real causal lift from naive correlation. Containerized with Docker and deployed via GitHub Actions CI/CD to an interactive Streamlit dashboard.