I get a company's operation to run on its own.

Lisbon, at 22

My first time abroad was Portugal. I went through Ciência sem Fronteiras (Science without Borders) to Instituto Superior Técnico in Lisbon, on a mobility placement in the integrated master's in mechanical engineering, and completed six course units, among them energy in transport, energy management and electrical systems.

What stayed with me from that year was not the credits. It was joining the Transport, Energy and Environment research team, working on the energy and dynamic characterisation of electric vehicles, with on-road monitoring that compared their energy and environmental performance against conventional vehicles.

In other words, at 22 I was already measuring energy consumption in real conditions, with instruments, to compare it with what theory predicted. Measured work against assumed work. It is what I still do today.

Beijing, after that

Professors and coordinators of the engineering programme at UnB nominated me to the Brazilian Space Agency, I went through their screening and was selected by the Chinese government scholarship programme. The destination was Beihang, and the subject became aerospace science and technology, with research on microsatellites.

This time it was not only a different country. It was another alphabet, another way of teaching and a field I did not know. I finished with the best thesis defence of my class and with work presented at the Global Space Exploration Conference.

I took part in the BUSAT-1 programme, Beihang's 2U CubeSat in the QB50 constellation, launched in April 2017. My master's was the thermal control architecture: I simulated the orbital environment to size heat dissipation, heaters and blankets, so the electronics would work within the temperature range they can withstand.

A satellite has no maintenance. After launch nobody goes up to adjust anything, and the margin you did not put into the design no longer exists. That is where I learned to leave margin for what can go wrong, and it is the same rule I apply today to a system that has to run with nobody watching.

Why I started building

It was not a calling for software. It was a need for process.

I ran processes that were designed correctly, and the results still varied. A good process, carried out by different people on different days, under different pressure, does not come out the same. And the part that varies most is usually the most repetitive one, which is exactly the part that wears out the person doing it.

So I started automating to take weight off the people who were there. Training a system takes roughly the same time as training a person, sometimes less. The difference is that once it has learned, it repeats the same instruction the thousandth time with the same accuracy as the first. The person is free to do what no machine does.

The mistake it took me years to see

I built systems for other people's operations with a discipline I did not apply to myself. Every project of mine has a decision log, checks that fail the work, and a document that states what is live. None of that care went into the record of my own work.

The result is that a good part of my engineering became invisible. Substation design, protection studies, tax reports and public tenders were on no CV, no profile, nowhere. They existed only in the delivered work and in the memory of whoever was there.

This page is the correction of that mistake, and it took far too long to exist.

What I am building

I have an autistic daughter, support level 2, non-verbal. What I build today is her future, and it will go on being that for as long as it takes.

That answers a question this page raises without explaining. I insist on systems that run on their own, on checks that fail the work before the error gets out, on documents that tell the truth about their own code, and on a public repository called não depende de lembrar ("doesn't depend on remembering"). It is not an aesthetic preference. I learned, in the place where it matters most, that anything that needs someone present in order to work will fail at some point.