Senior ML engineer with seven years of shipping production AI in education and defense. I write interactive explainers for the technical foundations behind modern computing — machine learning, statistics, and the systems beneath them.
I'm applying to the Doctor of Health Informatics program at UTHealth McWilliams for August 2027, and I'm interested in the intersection of large language models and clinical data. Open to senior ML / applied-research roles in parallel — see /now for what I'm actively working on, or get in touch below.
Full-stack capabilities from ML engineering to cloud architecture, with a focus on shipping production systems.
Building production ML systems that solve real-world problems
Designing scalable data systems from concept to production
End-to-end development from backend APIs to interactive frontends
Driving technical excellence and business outcomes
I haven't given a public talk yet — but I'd like to. If you organize a meetup, conference, or seminar, here are talks I'm ready to deliver.
Most explanations of how a CPU works either stop at handwavy block diagrams or jump straight to assembly. I'll walk through how I built the interactive transistor and logic-gate simulators behind my 'Silicon to Systems' series, what makes a technical explainer click vs. fall flat, and the React/Canvas patterns I keep reaching for. Target audience: anyone who teaches CS, writes technical docs, or wants to see what's possible with a static site and good intentions.
A field report from rebuilding a multi-modal LLM content pipeline (slideshows, podcasts, quizzes) from a working Go POC into production Rust. What we found, what we'd do differently, and where Rust paid for itself vs. where it didn't. Frank about the costs — Rust isn't free.
Building VoltVault meant ingesting the NFPA-70 National Electrical Code — 1000+ pages of tables, definitions, and references — into a semantic-search system that licensed electricians would actually trust. I'll walk through the chunking and retrieval choices, the failure modes specific to regulatory text, and what transfers to other compliance-heavy domains (healthcare, finance, legal).
A look inside an LLM-backed tutoring system serving real students at a regulated education company, framed through the lenses of personalization, latency, and safety. The clinical-trial-adjacent constraints (no hallucinated medication doses, no leaked PII, defensible content) and how the architecture was shaped by them.
Organizing an event? I'd love to talk — drop me a line.
I'm always interested in discussing new projects, creative ideas, or opportunities to be part of your vision.