I am currently pursuing a PhD in the University of Marburg. My research is on Large Language Model evaluation and impact, focusing on aligning downstream tasks and end-user application success, by using interpretability techniques to investigate failure modes of both LLMs and benchmarks.

I have an unrelated bachelors in Mechatronics from Puc-Minas, a related bachelors in Linguistics from UFMG in Brasil. Also a masters in Language Science and Technology at the University of Saarland. I’ve previously worked on semantic models for legal text, developing automatic essay evaluation, modeling language change with word embeddings, curating datasets for LLM training, developing pipelines for dataset creation and evaluating language models for hallucinations.

My current research interestes are:

• LLM Evaluation and Benchmarking
• Interpretability for LLMs and Benchmarks
• AI Impact evaluation
• Hallucination Detection and Evaluation