A one-person studio. Shipping AI-native software end to end.
Fifteen years of shipping software, now spent on the products I want to exist.
I've been a software engineer across e-commerce platforms, internal tools, and consumer products, and I've watched the cost of building software collapse over the last two years. Oqodo is what I'm doing with that. A small studio out of Minneapolis where AI orchestration is the actual craft, not a marketing claim.
Before this: senior engineering roles at Twin Cities companies including Evereve, plus side projects across music, writing, and local civic software. The studio is the resume now, and the resume is on the homepage.
Six things we hold on to when the work gets loud.
Orchestration, not vibes
The interesting craft right now is orchestrating AI to ship real software. Not throwaway prototypes. Production systems that solve real problems for real users.
Studio as the resume
Oqodo is the studio where that craft becomes products. We build what we want to exist, deploy it on infrastructure we control, and let the work be the resume.
Ship the products we want
Every product on this site started as a personal problem. We won't ship something we wouldn't pay for or use ourselves.
End to end, in house
Design, code, infrastructure, support. One studio, one stack. The seams are ours to own; the bugs are ours to fix.
Public when it's worth it
Some products are open source. Some are quiet experiments. We share the work when sharing makes it better, not as performance.
Slow on purpose
Thirteen products is the upper bound, not the goal. Most weeks are about depth on one or two.
Roughly the path from an annoyance to a public URL.
A problem we keep returning to
A product starts as a recurring annoyance, usually our own. We let it sit for a few weeks. If it's still there, it's real.
Sketch in Claude Code
First passes happen as live, throwaway prototypes. Cheap to abandon, cheap to fork. We're looking for the shape of the thing.
Choose a stack we already operate
Next.js on Cloudflare for most things. Rails 7 when the data model is the product. We don't introduce new infrastructure unless it earns its keep.
Ship a thin version, then watch
The first public version is deliberately incomplete. We watch where real users get stuck, then decide whether to deepen it or move on.
Maintain or sunset, never drift
Products either get continued investment or move to the open-source archive shelf. Nothing rots in production unattended.
The boring, durable tools we lean on to build the not-boring things.
What's on the bench this week.
UPDATED · 2026-05-28
Earshot. Early design conversations with prospective design partners.
Sprintbook. Turning a consulting workflow into a standalone product.
Moxmo. Live with paying users; building the v2 behavioral layer.
recursive-ai. Open-sourced pattern, public writeup in progress.