I build products around AI, developer tools, and coding-agent workflows.
I spent about a decade in frontend and product engineering before leaving my Lead Frontend Engineer role to work full time on my own products.
I write essentially none of the implementation code now. Coding agents do that.
My work has moved toward deciding what should exist, writing and refining specifications, connecting the right context, evaluating the work, breaking the product, measuring the result, and deciding what ships.
Product intelligence for agentic builders.
ShipFoundry tracks changes across the tools behind a product, checks what matters against the product and codebase, and turns useful changes into evidence-backed work for coding agents.
A global platform for scuba, freediving, and snorkeling with more than 12,000 dive spots and 234 destination guides.
DiveJourney is also my largest proving ground for building and operating a production product almost entirely through coding agents.
I'm directing and dogfooding a private local control plane for several coding agents working on one initiative together.
The core bet is:
One initiative owns the branch. Agents own temporary work.
Instead of giving every agent an isolated branch and merging the results afterward, Riata coordinates agents inside one shared branch and worktree.
The current system includes durable sessions, exact-path leases, actor-attributed brokered commits, isolated agent runtimes, commit-bound review, durable findings, and a final independent branch review.
The newer task layer connects specifications, acceptance criteria, expected paths, and multiple agent seats to one durable unit of work.
It is still an experiment. The larger system is not finished, and I have not yet proven that it beats a strong single-agent workflow on real initiatives.
The repository remains private while I dogfood that question.
I write about what actually happens while working this way: coding agents, developer tools, technical ownership, product building, performance, distribution, and the parts that fail.
Start here:
Rebuilding the Machine to Keep Up With the Fleet
I'm interested in serious remote work with U.S.-based, globally distributed companies working on AI products, developer tools, developer experience, developer education and relations, technical product, coding-agent workflows, evaluation, and adjacent problems.