At Apart Labs, I researched how large language models can infer human preferences as judges and routers. I co-authored a framework that learns an interpretable aggregator over multiple rubric-conditioned LLM judges, presented at the NeurIPS 2025 Reliable ML Workshop.
At the University of SãoPaulo (USP), my M.Sc. research focuses on database design and knowledge graphs of Brazilian literature for querying with LLMs. I also built recurrent neural network models to forecast regional soybean yields from historical yield and climate data.
As a Teaching Assistant (Monitor) at the University of SãoPaulo (USP), I teach Principles of Algorithm Development and support students with programming, algorithms, and data structures. I previously supported the development of the Latin American Network for AI Safety at CEEALAR.
I work across machine learning, LLM evaluation and routing, knowledge graphs, and time-series forecasting. I’ve also contributed to the Linux kernel and GNOME, and built an internet-connected micro weather station as my technical diploma final project.
