Portfolio / Build notes / Books / LinkedIn / Say hello
I'm a staff-level full-stack engineer, product builder, and author in Colorado. I've spent 14+ years building software and leading engineering teams. I still like being close enough to the code to follow a weird bug from the interface to the database.
Next.js, TypeScript, and applied AI are my home base. I care about how a product feels, what happens when it breaks, and whether it actually makes someone's day better. That curiosity has turned into agent tools, native apps, accessibility projects, and a couple of books.
|
A workspace for building and supervising AI work. A web product and a local-first Mac app, with explicit approvals and inspectable runs.
|
A consent-first Mac meeting coach. Local transcription, opt-in AI suggestions, and a clear boundary: you always speak for yourself.
Source & local install · Early release |
|
Help an agent find the capability it's missing. Evidence-backed routing across MCP, A2A, and REST, without holding provider credentials.
|
Free, guided AI learning for people with Down syndrome and other intellectual disabilities. Patient steps, accessible interactions, no required account.
|
|
A little more peace of mind when meeting someone new. Private date plans, trusted circles, timed check-ins, and quiet alerts.
|
A React component library with a brutalist streak. Bold visuals, keyboard-friendly interactions, and accessibility built into the components.
|
I like sharing the decisions and working code, not just the finished screen.
| Start here | What you can dig into |
|---|---|
| Handrail | A computer-use reference implementation: discover once, replay without a model, hand control to a human in the same session. Synthetic demo, real browser execution. |
| AI System Architecture Blueprints | Architecture diagrams, decisions, threat models, and cost checks. The reasoning alongside the design. |
| Release Safety Team | A supervised delivery reference with an offline sample, explicit approval boundaries, and reviewable run receipts. |
More tools, experiments, and things I've built
- Agent Team Starter: runnable TypeScript agent teams with approvals and evals.
- Browser QA Agent: guarded Playwright QA with screenshots and runtime evidence.
- GitHub Repo Maintenance Agent: issue-to-patch automation with path limits, checks, and human approval.
- Workflow Check: a zero-dependency validator and GitHub Action for agent workflow contracts.
- Agent Team Examples: reviewed workflows for releases, incidents, refunds, leads, and research.
- Ember Coast: a full-stack luxury travel planning product. Build story.
- ViralVault: a builder-focused brief on AI, infrastructure, and security shifts.
- stdout: autonomous developer stories generated and published from local hardware.
- vault-kalshi: a forward-testing research lab with reproducible data and published negative results. Research, not a claim of a profitable strategy.
One book about helping people do good work. One about getting useful work out of AI.
| People & leadership | AI & craft |
|---|---|
| The Plant Was Never the Problem | From Prompt to Proof |
| Creating the conditions for people to speak up, grow, and ship. | Using AI to think clearly, finish real work, and verify what matters. |
- Own the whole problem. Product decisions, frontend, APIs, data, deployment, and the rough edges after launch.
- Keep people in control. Consent, understandable permissions, accessible interfaces, and a way to stop.
- Make the system explain itself. Clear state, useful logs, reproducible tests, and failure modes you can reason about.
- Help the next person. Write down the tradeoffs, share what I learn, and leave the code easier to work in.
Next.js TypeScript React Node.js
PostgreSQL AI agents MCP Playwright
Swift Rust Expo
Still curious. Still building. Occasionally writing a book about it.