Data Science applied to games: my internship at Wildlife Studios

Why the hard part isn't writing the query, but figuring out which question is worth answering.

By Gabriel Lucena, Data Science Intern · September 2026

Before Wildlife Studios, I was an applied mathematics student at Unicamp with experience developing fan versions of old-school games. Building those taught me how games actually work under the hood, but I also wanted to work with data. At Wildlife, I had the opportunity to join both worlds and work with Data Science applied to games.

Working with data means understanding how players interact with the game

I joined early last year, on the New Games Insights team. I was expecting to explore the data, code a few queries, and generate charts to answer questions, but I quickly realized the scope was way broader than that. For each simple question — “why are players dropping at this level?” — I had to understand what we wanted to measure, why we wanted to measure it, and how I should answer it. It is really important to understand whether we are answering the right question. It matters just as much to know which outputs and findings are the most valuable. Most of the work happens before the code: understanding the problem, and discussing it with the team until we get there. It is slower and far more collaborative than I imagined, and that is exactly what makes it exciting.

My job is to understand how the game actually works: what makes it fun, what makes players play one more time, what makes players spend money. We also work to understand what is working and what is not. For example, it is common to explore the game’s calibration and understand when its economy is broken — and why it is broken, so we can act on it.

One of my first analyses was exactly that. Players were accumulating a huge amount of currencies. I looked at the currency balance across levels and the accumulation was clear. Then I broke it down into how much currency players were earning and spending: spending was low and steady from level to level, while earning kept growing. The proposal was to calibrate the mechanic that was exponentially increasing player income, and to keep discussing what opportunities we had to increase spending through progression. The finding was clear enough that fixing the economy calibration was prioritized on the roadmap.

While working on the New Games Insights team, I got to work on several game prototypes we were testing and developing to improve each game as a product. It is really cool to deep dive into several games: we can compare how different games work and improve our own product, and sometimes we need to really understand one game deeply and propose creative ways of breaking down a problem. Today I work on two games in more depth. Passing through several games first gave me repertoire: each game has a different kind of problem, and after a few of them you start recognizing the patterns.

From a hackathon to a real project

From time to time, we step away from our main activities and work on a hackathon, where we can build an entirely new game in three days, or work on another kind of project. In my first hackathon, I chose to work on a bot that could answer open-ended questions about player behavior, so we could spend our time on the more time-consuming questions and empower product managers to get answers outside dashboards faster.

The idea and its potential were good, and we decided to continue working on the project after the hackathon. It became my main project for a while: the bot ran in our Slack, and we had jobs running in production. But in the end, adoption did not hold up enough to justify prioritizing it — so we decided to stop investing in the project.

In practice, it did not work out the way we expected, and we moved on to the next game project. But while developing it, I got to broaden my knowledge and learn how to break an open question into three parts: what we want to measure, why we want to measure it, and how to answer it — turning a vague question into a closed, reliable and rigorous analysis. Those patterns made me better at the games I worked on later, and they also helped me improve our libraries and tools — the ones that let us build the most common analyses faster, and with more confidence. In the end, the process of working on this project was super important to how I work today.

AI helping to deep dive into analyses

Those patterns became the workflow I use every day. I structured a way of working where the agentic AI takes care of the operational part of the analysis: writing code, building notebooks, generating dashboards. I keep the part that really needs judgment: understanding the problem, questioning the hypotheses, and deciding what the analysis has to prove before I believe it.

Day to day, operational work is not the bottleneck anymore. An analysis that used to take days now flows much faster, and with more consistency, because the process is standardized and reproducible. The libraries and tools we built together help with that consistency: they give us standard functions and conventions we agreed on as a team, so we can trust what we are producing. That leaves me free to spend my time on the questions that are actually hard.

And it is not just me. I have seen the use of AI grow a lot in the team over the last year and a half. The question is not “is it worth using?” anymore, it is “how do we use it well?”.

The data culture I found here

Before joining, I did not really know what to expect: it was my first experience at a tech company. What I found here surprised me — decisions really guided by data, and deep technical discussion. Data is not a bunch of reports someone sometimes asks for at the end of the process: it is part of how decisions are made from the beginning.

But what impressed me the most was the autonomy. As an intern, I was expecting to receive well-defined tasks and execute them. Instead, I am part of the discussion from the start, and I get to help guide prioritization. And the team genuinely wants me to get the most out of this experience.

During this period, I had two managers, and both of them gave me their time and backed me with consistent opportunities. They were always close, bringing me challenges that helped me learn, and they encouraged me to question things, to understand why we were doing them, and to share my opinion. “Why are we doing it this way?” was never a bad question here — it was an expected one.

The process is the fun part

What counts here is constant growth and learning. You have to think, question, test, and experiment until you get to the best idea. Repeat until it gets better. Over time, it gets easier to see progress in small steps.

At Wildlife, we learn that the best results do not come from complicated methods or processes, but from a team that pulls together. It was this environment that showed me we can develop ourselves while helping each other grow.

Get into it without fear, fail fast, adjust, keep going. The idea of never being done is what keeps me moving, and here — inside this culture — I am sure I will keep growing. This is a good place to grow.