At V-Lab / UFPE, I develop AVAMEC Interativo, the official MEC platform built on Redu. I created course-entry and course-completion questionnaires integrated with official IBGE and INEP data; reports from these features have reached about 200,000 users.
On the SDPA project, I developed Python and Flask REST APIs connecting a school-dropout risk model to dashboards for school managers. The XGBoost model I built with SMOTE achieved 0.912 accuracy and a 0.718 F1 score; this work became my undergraduate thesis and a first-author paper published at SBIE 2026.
I also built edital-rag, an open-source project that answers questions about public notices with verifiable citations, using hybrid vector and BM25 search. I'm developing DeixaQueEuFaço, a Django and React household-task PWA, with REST APIs and CI through GitHub Actions.

