

We are excited to welcome Hamidreza Amirzadeh to INSAIT as a doctoral student in Natural Language Processing (NLP), joining Prof. Iryna Gurevych’s research group!
His research focuses on mechanistic interpretability, representation learning, Transformers, and trustworthy NLP.
Hamidreza holds an M.Sc. degree in Artificial Intelligence, Computer engineering from Sharif University of Technology, as well as a B.Sc. from Amirkabir University of Technology – Tehran Polytechnic.
Hamidreza has authored and co-authored research papers published at leading venues, including EMNLP, COLING, and SemEval. His work explores how language models process contextual information, generalize to longer sequences, and improve their reliability, including through research on contextual biases, positional encoding, and also the detection of hallucinations in large language models.
Prior to joining INSAIT, he worked on machine translation explainability and quantifying token attribution for generating each token during translation as part of his Master’s thesis at Sharif University. He also gained further research experience through collaborations with the IT-Universitetet i København and Tilburg University. In addition, he has diverse industry experience, including DevOps and AI platform engineering roles.