About

I’m a senior frontend engineer with 20+ years of software engineering experience, including 12+ years building React products at Pearson Learning Solutions, where I lead frontend architecture and delivery. I’m comfortable across the stack — React, React Native, and TypeScript on the front end; Node.js, PostgreSQL, and MySQL on the back end; Docker (with Compose) to tie it together; and AWS and Google Cloud for deployment. The other thread in my recent work is hands-on AI engineering — LLM tool calling and embeddings in Python — both in side projects and in how I ship frontend code day to day.

Experience Link to heading

Pearson Learning Solutions — Senior Frontend Engineer (Software Engineer III) · 2014–Present

I lead frontend implementation on a five-engineer product team — breaking features into tasks, estimating them, and guiding them through delivery. I own the React/Redux architecture and coding standards for Live Response, Pearson’s real-time classroom assessment app used by more than 1,000 higher-ed students and instructors. That’s meant designing both instructor and student workflows: classroom geofencing enforcement, login flows and authorization-aware UI, and a client polling mechanism for near-real-time graded-question updates. Along the way I mentor two engineers and partner directly with a UX designer to turn Figma designs into responsive, accessible React.

Lately I’ve been wiring AI into that workflow: repository-level architecture guidance in AGENTS.md, CLAUDE.md, and SPEC.md, plus authenticated Figma MCP access in Claude Code — so AI-generated components come out matching the app’s React/Redux conventions instead of fighting them.

Earlier projects at Pearson: Freehand Grader, where I integrated a third-party math editor and routed its MathML output to Pearson’s scoring engine, and Learning Catalytics, where I built a background Rails notification service and contributed to its jQuery frontend.

Before Pearson — I architected and built Ruby on Rails applications as a consultant at CabForward in Austin (2012–2014), did contract Rails work for startups like Pago Mobile and iLoop Mobile as well as larger shops like Home Depot (2008–2011), and wrote Java at Drillinginfo (2004–2006).

Side Projects Link to heading

I recently built playcall, a natural-language NFL stats chatbot — ask a question in plain English and it answers from the nflverse play-by-play data in DuckDB, with every answer auditable. The source is on GitHub.

I also built Tunearby, a ground-up rebuild of my old Radio Stations Near Me app — the original written by hand before AI coding assistants existed, the rebuild done with an AI-assisted workflow end to end. The product: every AM/FM station broadcasting near you, sorted by distance — no account, no ads, no backend. The full FCC catalog ships in an on-device SQLite database, so it works offline. It’s React Native + Expo on iOS and Android, with a pure-React web version at tunearby.com that queries the same catalog client-side via wasm SQLite.

Speaking Link to heading

I’ve spoken at ReactDenver on “Building Android and iOS apps using React Native and Expo.”

AI & Tools Link to heading

Most of my recent work is hands-on AI engineering. Building playcall taught me how LLM tool calling actually works: the model runs a multi-turn agent loop — picking tools, writing SQL, reading results back, and self-correcting on errors. I implemented that loop across three providers (local Ollama, OpenAI, and Anthropic), streamed it to the React front end over server-sent events, and kept it honest with an eval suite of ground-truth test cases. My throughline for AI: don’t trust the model, trust the verifier.

I’ve worked with embeddings and vector databases — turning both text and images into vectors for semantic search and similarity.

Day to day, I use Claude and Codex, and run local LLMs with Ollama (qwen3-coder for code completion). I also work in Python — it’s been a great fit for AI and API work.

Cloud & Deployment Link to heading

I deploy and run full-stack apps in production on AWS and Google Cloud. I shipped fadethepublic.com to Google Cloud Run as two separate services — a static Next.js site served by Nginx and a FastAPI API — each containerized, mapped to its own custom domain, backed by Firestore, and set to scale to zero when idle. The full walkthrough is in Deploying a Full-Stack App to Google Cloud Run.

Education Link to heading

  • MBA, Finance — University of Texas at Austin, McCombs School of Business, 2006
  • B.S. Computer Science — University of Texas at Austin, 1998

When I’m Not Coding Link to heading

You can usually find me fly fishing, skiing, training Muay Thai, or at the poker table.

Elsewhere Link to heading