I'm an Information Systems student from Rio de Janeiro. I work mostly on the backend: Python, APIs and databases.
Lately I've been building things with LLMs on top of real data. Public data is usually a mess, so most of the work happens before the model gets involved. I learn by building projects and finding out what breaks.
- Building — an API that pulls official government gazettes, extracts structured data from raw text and answers questions about it through an LLM
- Learning — async, dependency injection and testing in FastAPI
- Exploring — LLMs, RAG and vector databases
- Goal — my first role in applied AI or data-heavy backend
diario-oficial-rag · in progress
Question answering over Brazilian official gazettes. The pipeline collects raw publications, pulls structured fields out of them with regex, indexes the text in a vector database and answers questions through an LLM, citing the source publication.
Python · FastAPI · Qdrant · Docker
Full Minesweeper implementation in Java with a Swing interface, multiple difficulty modes, match history and a ranking system.
Java · Swing · OOP