Ricardo Rei is the Head of AI Research at Sword, where he leads a team developing AI for healthcare, focusing strongly on LLMs for Mental Health and physical pain. Previously, he was a Senior Research Scientist at Unbabel, leading the post-training of Tower LLM and supporting the deployment of these models across research, product, and sales. He was also a key contributor to both Tower LLM and EuroLLM, working on areas ranging from pretraining data quality to evaluation. Earlier in his career he specialized in model evaluation, serving as the principal developer of COMET, now widely used as the standard for machine translation evaluation. His PhD thesis on evaluation earned the EAMT Best European Thesis Award in 2024.