Thomas Vakili

Thomas Vakili is a doctoral researcher at the Department of Computer and Systems Science, Stockholm University, affiliated with the Natural Language Processing Research Group. His work intersects language technology, artificial intelligence, and privacy protection, focusing on mitigating data leakage risks in large language models while maintaining utility.

• Education: MSc in Computer Science and Engineering (KTH Royal Institute of Technology)

Research explores privacy-preserving NLP through techniques like pseudonymization and de-identification, with applications in clinical data and Swedish electronic health records. Recent projects include End-to-End Pseudonymization of BERT models and analysis of eponyms in clinical text.

Publications in BMC Medical Informatics, LREC, and NoDaLiDa conferences address privacy risks, model utility, and cross-lingual NLP. He supervises bachelor's and master's theses and teaches courses in NLP, Internet Search Techniques, and Digital Business Strategies.