About
A small studio,
on purpose.
Who
Alex Jiménez
Fifteen years building software — SaaS platforms, healthcare systems, and AI products at startups. Healthcare is the part that shaped how we work: in regulated systems the boring things decide everything. Whether data moved correctly. Whether there is an audit trail. What happens at 3am when something is down.
Most of that work was for companies big enough to have engineering teams. Milpa Craft exists because of the conversation that kept happening at the edges of it — a smaller company with a real, expensive, unglamorous problem, too small to hire engineers and too specific for anything off the shelf. They get quoted like an enterprise, or handed to whoever is free. That gap is the whole business.
Based in Mexico City. Working US hours. Projects delivered in English.
The name
Milpa
A milpa is a field planted with corn, beans and squash together. The corn gives the beans something to climb. The beans put nitrogen back into the soil that feeds the corn. The squash spreads low and wide, shading out weeds and holding water in the ground.
None of the three is remarkable alone. Planted together they outproduce any of them grown separately, and they have done so in Mesoamerica for several thousand years.
That is the whole idea behind the work: systems where the parts hold each other up. Craft is the other half — built by hand, by people who will still be here when you need it changed.
How we work
Senior people only
Nobody learns on your project. The person who scopes the work is the person who builds it.
You own everything
Repository, cloud accounts, domains, documentation — in your name from the first commit. Leaving should be boring.
Small before large
The first phase is deliberately narrow, so you find out how we work before the budget is committed.
Plain language
You should be able to explain to your own team what we built and why. If you cannot, we have not finished explaining it.
Tools
The stack, briefly.
Python, JavaScript and TypeScript, Java, Postgres. We pick boring, well-understood tools, because the point is software you can still hire for in three years — not software that impresses another engineer.
Where AI earns its place — reading documents, classifying requests, pulling structure out of mess — we use it and show you the confidence thresholds. Where it does not, we say so.