Most of my work starts with data nobody collected for analysis: court case records, World Cup match data split across eras and sources, footage from a gym camera. My job is turning that into a table someone can trust. The court, the basketball league and the World Cup app are the same problem with a different mess.
I work on data modeling and the data warehouse behind the court's judicial statistics.
I write Python pipelines that call the Anthropic API to classify case texts and standardize inconsistent fields before they load into the warehouse.
I also support other courts working on their performance.
On the team building SpotAI, still in development. More under Selected work.
A UFPB event that puts students in a room with people who already work in data.
A computer vision pipeline that pulls per-player stats from video of Paraíba 3x3 basketball league games: shots, heat maps and distance covered.
A voice assistant that builds a résumé from a plain conversation, turning informal answers into professional wording for low-income workers.
My part was the RAG layer: retrieval over job-title vocabulary, résumé templates and job descriptions.
We spent 6-7 hours building inside Google's São Paulo office and took Paraíba to 4th place. Still one of the best days of my life.
A web app with every national team's World Cup record from 1930 to 2022, plus how close its style sits to other teams and eras.
Similarity uses only the features both sources share, and shows a confidence level with how many metrics went in. Data: StatsBomb Open Data (1970, 2018, 2022) and historical records compiled from public sources (1930 to 2014).
Text goes through spaCy and regex cleaning, TF-IDF with bigrams and a PassiveAggressiveClassifier, and comes out labeled REAL or FAKE. A sentiment pass runs on the side.
The first time I opened a terminal, I was following a YouTube tutorial to install something in Minecraft. It worked and I felt unstoppable.
Most of my family works in tech, and they're the people I look up to most, as engineers and as people.
Data came in my last year of high school, when a cousin finishing a data degree in São Paulo walked me through what he actually did all day. Research, investigation, an algorithm figuring things out from evidence: I was hooked on the spot.