A real-time satellite orbit tracker and collision predictor built with a custom C++ SGP4 propagation engine.

• ▹ C++ orbital engine compiled via pybind11 into Python, computing positions for hundreds of satellites in real time
• ▹ Full coordinate transform pipeline from TEME frame through ECEF to geodetic lat/lon/alt using NASA's SPICE toolkit
• ▹ FastAPI REST backend with 6 endpoints serving a Cesium.js interactive 3D globe with orbit trails and time controls
• ▹ 279 automated tests across 7 test files covering the C++ engine, coordinate transforms, and all 6 API endpoints

A machine learning pipeline predicting NFL player performance, achieving measurable outperformance against professional benchmarks.

• ▹ Built 40+ position-specific CatBoost models trained on 109 weeks of NFL data with 50 engineered features including Vegas lines and rolling variance metrics
• ▹ Achieved 4.26 MAE, outperforming professional DFS platforms (industry benchmark: 4.5–5.5 MAE)
• ▹ Built a Streamlit dashboard for exploring predictions and model insights

The site you're looking at right now. The background is a real-time star viewer rendering the night sky from where I'm based, Los Angeles. Click the star icon to explore it!

• ▹ Rendered 15,598 real stars from the HYG v42 catalog with B-V color index mapping, magnitude-based brightness, and per-star scintillation across 15 ES modules
• ▹ Computed planetary ephemeris at runtime using JPL Keplerian orbital elements, with Sun and Moon positions via simplified Meeus algorithms
• ▹ Built 88 IAU constellation overlays with interactive highlight, time controls with rewind, and object info popups
• ▹ Built a Python preprocessing pipeline combining HYG catalog, Stellarium constellation data, 27 DSOs, and Milky Way centerline into a 521 KB JSON file

A binary classification pipeline predicting cryptocurrency price direction by combining technical indicators with social media sentiment analysis.

• ▹ Built pipeline combining 5 technical indicators (MACD, RSI, Bollinger Bands, and others) with FinBERT sentiment scores from Reddit and X across 3 live data sources
• ▹ Integrated Kraken, Reddit, and X APIs for real-time ingestion; applied FinBERT transformer for financial sentiment classification

• ▹ Lead a team of 8+ musicians through weekly rehearsals and live performances
• ▹ Manage scheduling, set lists, and real-time coordination across weekly services and special events
• ▹ Coordinate with audio/visual tech volunteers to run live sound production

• ▹ Coordinated deliverables across multiple concurrent infrastructure projects
• ▹ Tracked project budgets and reconciled expenses against allocations
• ▹ Managed permit acquisition, regulatory compliance, and project closeout documentation

• ▹ Maintained detailed patient records, medication logs, and vital sign documentation on regular schedules
• ▹ Developed individualized care routines for patients with dementia and Alzheimer's, adapting to each patient's needs