From zero, December 2015
In December 2015 I co-founded a digital out-of-home media company in Brazil: screens in elevators and malls, selling time to advertisers. There was no technology yet, so I built it. The part I'm most proud of is the proof-of-play platform, because every ad play is billing evidence and the screens lose the network for hours. Playlogs are cached on the device, uploaded in batches to an ingestion server, then Kafka into ClickHouse.
It grew from the first screen to 15,000. Today it takes about 200 million proof-of-play events a day and keeps more than 300 billion rows in our own ClickHouse, the largest table past 100 billion. Some parts had to change on the way: when MySQL got to two billion rows, the store moved to ClickHouse on bare metal, and the fleet kept growing on top of it.
My partners raised the money and ran sales. That was a division of work. Building the technology and, at the same time, raising capital and selling is more than one person's day. I was CTO until the end of 2024 and stayed close to the commercial side the whole time; the occupancy screen in the cases is where the sales team starts the morning.
How I work
I usually come in by one of two doors. I build from zero, when a product has to exist by a date and then survive real use. Or I take over a system that is already running, with revenue on top of it, and that started to break because the business grew. I like both. When I build from zero I design for the second year, because I've seen what breaks there.
Either way I measure first: counts at the source, counts at the destination, and the difference between them. It takes a few days, and from there we discuss a number with a source.
I use AI every day, and a lot of what lands on my desk now was built fast with it. That's fine, it got the product to the demo. What's usually missing is the boring part that makes it hold: measurement, and a data model that survives the next year of growth.
Now
During 2025 and 2026 I kept leading technology projects at the company I co-founded, and I still do.
I'm also building from zero a performance platform for professional athletes, for a sports-science company in London. Python and FastAPI on the back, Next.js on the front, more than 300,000 data points pulled from a vendor portal that has no API, and the AI insights engine on top of that data is the next part, coming soon.
Outside work
I grew up in the Brazilian racing paddock, where my father is a legend. My habit of measuring everything comes from there, and also the respect for things that can't fail on race day.
I also train jiu-jitsu.
And sh4 means nothing. It's short and easy to remember, that's all.