About Ubbin Labs
Behind the work.
Ubbin Labs is the engineering team behind GFoundry. We have been building things since 1999, which is long enough to have strong opinions, a few scars, and one or two stories we are still not entirely sure how to tell.
We built Palco Principal from scratch – the largest Portuguese-language music social network, running across Portugal, Brazil, Angola, Mozambique, and Cape Verde. With 70,000 tracks and no Spotify anywhere in sight, we designed our own recommendation engine before the tooling existed. It worked. Then the 2008 financial crisis wiped out the online advertising market, and we learned something important about what happens when the engineering holds but the business model doesn’t.
When the App Store was still a novelty and most developers were still figuring out how to ship one native mobile app, we took on Heathrow Airport. Three platforms – iPhone, BlackBerry, Nokia QT – built from scratch, for one of the busiest airports in the world. There were nights that did not end. The apps shipped in January 2012. Heathrow used them.
We then ran a formal R&D programme with Fraunhofer Portugal, building stream mining infrastructure with real behavioural data, which became the technical foundation of GFoundry. Ten years later, that platform runs in 26 languages and is used by enterprise clients across multiple countries.
There were other projects. Many of them. We will spare you the full inventory — partly because this page would never end, and partly because the stories worth telling are already in the writing section. The rest we keep for the second drink.
Written by people.
Read articles by João Carvalho and Pedro Trindade in the journal.
Read the full site manifesto
We are the engineering team behind GFoundry. We have been building real products for over twenty years.
This is not a portfolio. It is not a pitch. There is nothing here to buy.
Ubbin Labs is where we write about what we built, how we built it, and what we learned — including what went wrong. We believe the decisions behind the code are more interesting than the code itself, and that honest engineering stories are rarer than they should be.
What we have built
The domain has changed three times. The problem has not.
In 2006, the challenge was 70,000 songs and 100,000 listeners, and the question was how to surface what each person wanted to hear next, without them having to ask. The answer was a recommendation engine built on collaborative filtering and behavioural signals, one of the first of its kind in Portugal.
In 2012, the question was how to understand the behaviour of people inside a social gaming platform: who they were, what motivated them, how they clustered. The answer was a stream mining infrastructure built in a formal R&D partnership with Fraunhofer Portugal, processing behavioural data in real time.
Today, the question is how to understand what is happening to people inside an organisation: who is disengaging, what each person needs to learn, what conversation a manager should be having. The answer is what we now call people intelligence.
Between those chapters: more projects than we will list here, and most of them not the kind that end up on a slide deck. Among them, native mobile apps for one of the world’s busiest airports, built across three platforms simultaneously by a team that had never done it before. A B2B SaaS platform that enterprise clients in multiple countries depend on every day. Products that shipped, and some that did not survive.
Small teams. Real constraints. No venture capital. The same obsession throughout.
What we are building now
We have been working in the people space long enough to be sceptical of what passes for AI in HR. Most of it is window dressing on top of existing data, or automation of things that were not worth doing in the first place.
What we are building is narrower and harder: people intelligence. Systems that help employees grow, that help managers understand what is actually happening in their teams, and that generate learning experiences and insights that did not exist before. Not generic. Not horizontal. Built specifically for the people domain, by a team that has spent more than two decades understanding how organisations work and where they get stuck.
Why this site exists
The patterns we discovered, the architecture decisions we debated, the failures we recovered from – none of that belongs to a product. It belongs to the engineers who lived it.
This site makes that visible.