Ricardo Rei leads AI research at Sword Health, where he directs work on large language models, multimodal AI, and the next generation of machine learning systems for healthcare. His research shapes the scientific foundation behind Sword’s AI Care platform, ensuring that the technologies delivering care to members are not only capable, but rigorously evaluated and trustworthy.
At Sword, Ricardo leads initiatives exploring how generative AI and language technologies can better support clinical workflows, personalize care at scale, and sustain the kind of engagement that traditional healthcare systems have struggled to achieve. He came to Sword as one of the field’s leading voices on how to evaluate AI systems honestly, and that discipline runs through everything his team builds.
Ricardo is the principal developer of COMET, now the standard metric for machine translation evaluation across the research community. His doctoral thesis earned the EAMT Best Thesis Award in 2024, selected as the best PhD thesis in machine translation in Europe that year. His research brings both scientific depth and practical discipline to Sword’s AI development, grounded in the conviction that powerful systems must also be measurably dependable ones.
EAMT Best Thesis Award, 2024
Doctoral thesis recognized as the best in machine translation research in Europe, awarded by the European Association for Machine...
Principal developer of COMET
Creator of COMET, the widely adopted standard metric for machine translation evaluation, with research citations in the thousands.
Education
Ricardo holds a PhD in Electrical and Computer Engineering from Instituto Superior Técnico, completed Summa cum laude in 2024 under the CMU-Portugal MAIA project, supervised by Professor Luisa Coheur at INESC-ID and Professor Alon Lavie at Carnegie Mellon University. He also holds a Master’s degree in Computer Science and Engineering from Instituto Superior Técnico, completed with Merit diploma and a grade of 18/20.
Experience
Before joining Sword Health, Ricardo spent over seven years at Unbabel, progressing from AI Researcher Intern to Senior Research Scientist. In that role he was one of the principal contributors to the Tower LLM and EuroLLM projects, working across pretraining data quality, post-training, and evaluation. He was also a Visiting Student at Carnegie Mellon University, where he worked with Professor Alon Lavie on machine translation evaluation, placing second at WMT shared tasks and winning the Quality Estimation shared task. Earlier, he served as an Early Stage Researcher at INESC-ID and as a Teaching Assistant at Instituto Superior Técnico for NLP and Logic Programming courses. At Sword, he leads AI research focused on building trustworthy, rigorously evaluated systems for healthcare.
