RESUME

• Architected a multimodal ML inference system in Rust — user-specific LoRA adapters across daily/weekly/monthly horizons — producing individual-level risk predictions from wearables, labs, vitals, nutrition, and behavioral signals.
• Built a counterfactual prediction engine (/what-if) for causal outcome estimates under hypothetical input perturbations, applying causal ML and decision theory to a production system.
• Designed a Temporal-orchestrated nightly personalization pipeline: per-user feature extraction, model inference, and intervention generation — not population averages.
• Scaled to 300+ person waitlist; incorporated as a Delaware C-Corp with HIPAA-compliant Postgres/Redis infrastructure from day one.

• Lead 6,000+ members across 126+ chapters in 30+ states and 4 countries delivering financial literacy and investment education.
• $50,000+ fundraised with Chipotle and Intertwined; Stock Market Competition awarding scholarships to 50+ participants; 3+ guest lectures from industry leaders incl. Wells Fargo VP.

• Co-founded student-led nonprofit; grew program to serve 20+ middle schoolers across 4 weeks of structured ML curriculum.
• Designed GreenSight: guided students from 0 coding experience through a full ML pipeline — Python fundamentals to predictive modeling to civic advocacy; students built a model identifying 15+ Bay Area high-CO₂ locations submitted to Sunnyvale city government; in talks with Mercury News for a feature.

• Built analytics platform integrating YouTube & Twitter (X) APIs, monitoring 10+ channel metrics and competitor benchmarks in real time.
• Created 5+ interactive Streamlit/Plotly dashboards; integrated Google Gemini AI for natural language querying, reducing manual reporting by ~60%.

• Bought, restored, and resold 10+ damaged vehicles and motorcycles; engine rebuilds, electrical troubleshooting, bodywork, and welding. Generated $10K revenue in 3 months.

Quant pipeline extracting timestamped trading signals from SEC filings (10-K, 10-Q, 8-K) via LLMs — grounded in Cohen, Malloy & Nguyen (2020); 4 engines: supply chain graph, risk factor diff, M&A detector, tone tracker; outputs to Parquet/DuckDB with FastAPI + webhook support.

Differentiable physics library (classical mechanics, FDTD EM, quantum, statistical mechanics) — fully differentiable end-to-end; enables inverse problem solving via autodiff.