First production deploy
A bike-share app in Porto. It stayed up all weekend, which felt like a miracle at the time.
[ whoami ]
Eleven years later most of what I know came from being on call for my own mistakes. That is the whole method: read first, ship something small, stay long enough to fix it.
// the desk, most mornings — Lisbon, GMT+0
eleven years, two clients at a time
[ principles ]
Held long enough to have cost me work, which is the only test that counts.
01 / read_first
I read the repository, run the build on a cold machine and write down what I find. Clients hate this for about four days and then quote the document back to me for a year.
02 / boring_wins
Three years on call for a payments API taught me what an interesting architecture costs at four in the morning. I will always argue for the dull option that a stranger can debug.
03 / access_is_spec
Keyboard order and focus states are part of the component, not a ticket for later. Retrofitting a checkout for an audit costs roughly six times what building it right cost.
04 / leave_a_maintainer
Every engagement ends with a walkthrough recording and somebody on your team who can change the thing I built. tsflow has three maintainers now and none of them is me.
[ log --reverse ]
Eleven years, five things worth keeping. Drag the rail.
[ currently ]
Four days a week on a payments platform in Lisbon, one day for whoever is paying for it. When that day is booked out I say so here rather than in an email three weeks later.
[ Facts ]