What I've shipped
Claude Code running inside a Tesla
Mac mini at home → WebSocket relay → in-car browser. I deploy features to production from the driver's seat by typing (or voice-prompting) the feature into my dashboard.
Superchargers tracked daily
Full-stack data pipeline: public JSON → SQLite → Fly.io-deployed dashboard. Globe visualization, charging-desert analysis, network timelapse — refreshed nightly.
Reproducible bugs found in Tesla's open-source Go code
Coverage-guided fuzzing (testing.F) + Semgrep static audit. Each finding shipped with a runnable reproducer. Responsibly disclosed.
v0 of TeslaPulse, end-to-end
Dashboard, data collectors, agent bridge, deployment pipeline — from blank repo to live site. Shipping speed matters; I can move at startup tempo.
The site you're on is the portfolio. Every card is live data; every signal has a working collector behind it. Poke around — that's the evidence.
Three ways to work together
AI Workflow Audit
90-minute call on your current AI/dev stack, plus a written report with prioritized recommendations.
- Deep-dive on your current Claude Code / agent setup
- Where the quick wins are vs. where you're over-engineering
- Written report delivered within 48 hours
- Bookable this week
Custom Build
You bring the idea; I ship the product in 2–3 weeks. Data pipeline, UI, deploy, docs — end-to-end.
- Public-data dashboards like TeslaPulse, for your industry
- Claude-powered agents with live tool use
- ETL pipelines with scheduled collectors
- Full source hand-off or hosted-for-you
Embedded Engineer
Dedicated AI engineer on your team. One-week trial first so we both get to know fit.
- Full-time contract or fractional (2–3 days/week)
- Claude Code workflows, agent architecture, production systems
- Slack-embedded · your tools, your sprints
- References on request
What I work on
Claude Code & agent workflows
Building agents with tool use, prompt caching, thinking, structured output. Production-ready agent loops, not demos.
Data pipelines & dashboards
Public-data aggregation, scheduled collectors, visualizations. Stack: Node / Express / SQLite / Fly.io. Globe.gl, D3, custom charts.
Security research
Coverage-guided fuzzing (Go testing.F), Semgrep with composed rulesets, govulncheck reachability, CI/CD posture audit.
Full-stack AI-native products
Backend, frontend, infra, content — solo end-to-end shipping. When your team is small and "just ship it" is the requirement.
Claude API migration & tuning
Moving from OpenAI → Claude, or upgrading Claude versions. Prompt caching, cache-hit analysis, cost optimization.
Technical writing
PSIRT-grade disclosure reports, engineering documentation, product-spec writing. I can ghost-write your engineering blog at publishable quality.
Ready to talk? So am I.
Quickest path is an email with what you're trying to build. I reply within one business day — usually same-day for inbounds that include a clear problem to solve.