Senior full-stack engineer
I turn messy operations into software that doesn’t need babysitting.
Import pipelines, internal tools, and AI-assisted workflow automation for operations that outgrew their spreadsheets.
TypeScript / Python / Next.js / FastAPI / PostgreSQL / AWS
01 — Featured
What I’m shipping right now.
Shipping nowMac app · MCP server
ScreenBlurt
What it is
A Mac menu-bar app that reads text aloud in a natural voice. Draw a box around any part of the screen, including text in images, or highlight a selection, and hear it. It also ships an MCP server, so Claude Code, Codex, or Cursor can speak up when they finish or need a decision.
Why I built it
I could talk to Claude Code, but it couldn't talk back. Its built-in read-aloud spoke every tool call, so I built a tool to read only the parts I selected. Then I realized I could give Claude Code an MCP server and let it speak for itself. Working with it started to feel like working with a voice agent.
Swift · Next.js · TypeScript · PostgreSQL · Stripe · OpenAI TTS
02 — Selected work
One built by hand. One built with as much leverage as I could responsibly throw at it.
Where Is Justin is the fundamentals project; RigPlan is agentic product operations running in production — together they’re a fair picture of how I use tools: pick the right ones, keep your taste, and make the system earn its keep.
AI-leveragedProduct operations
RigPlan
A founder-operated planning platform for camper vans, RVs, trailers, and tiny homes. Behind the product, AI-assisted catalog workflows import, rewrite, categorize, and QA-check thousands of affiliate products so one person can keep the whole operation moving.
Next.js · TypeScript · Vercel AI SDK · Firecrawl
Hand-builtFundamentals
Where Is Justin
A map-first travel log built without AI assistance, on purpose: Python/FastAPI backend, PostgreSQL data model, and a separate Next.js frontend. The point was fundamentals end to end — schema, APIs, publishing flow, map UX — and it stuck around because it became a useful little system for shipping trip notes.
Python · FastAPI · PostgreSQL · Next.js
03 — Client work
Most of what I’ve built runs inside someone else’s business.
The products above are mine, so I can show them. The rest of the decade went to working embedded with client and product teams. The client names stay private. The outcomes don’t have to.
Serverless · AWS
A photo pipeline moving up to 20,000 photos a day
Carries photos from photographers across the country into a publishing client’s CMS and fills in the records from photo metadata. I designed it, built it, and run it on my own. It is still in production.
Next.js · Cost
A caching bug that was costing thousands of dollars a month
A middleware bug had silently turned off caching across a media client’s entire site. I traced it and fixed it, and the unnecessary API spend went away.
Event-driven · AWS
One event system feeding every downstream consumer
Enriches CMS content events with canonical IDs and fans them out over SNS to the GraphQL API behind the client’s web and mobile apps, a legacy API, and the analytics warehouse.
Angular · Testing
Greenfield apps, from untested to near-complete coverage
Built several new applications on a three-developer team, owning implementation and much of the UX design, and took the client’s frontend from no tests to near-complete unit and E2E coverage.
Live commerce · Team lead
Founding engineer, then team lead, on Videeo
CommentSold’s white-label live-video-commerce product, still on the market. I built its embeddable SDK and stream player, then led the team: four engineers and a QA specialist.
Full history on the résumé →
04 — Field notes
Notes that survived the work.
AI-assisted operations, framework behavior, build decisions, and the gotchas that only show up in production.