Yonatan Belinkov
Yonatan Belinkov researches language-model interpretability, robustness and control. His career connects language study with analysis of learned representations. Belinkov studied mathematics and Arabic at Tel Aviv University as an undergraduate, then completed a master’s thesis on an Arabic dialect. He also worked as an IntuView software engineer. Belinkov completed his MIT doctorate under James Glass. His thesis examined internal language representations in deep-learning models. He later worked as a Harvard postdoctoral researcher with Stuart Shieber.
Timeline
-
2022
Locating facts inside GPT
Belinkov coauthored the NeurIPS paper Locating and Editing Factual Associations in GPT.
-
2025–2026
Research across institutions
He spent a year at the Kempner Institute in 2025–2026 and is an associate professor at Technion. His research includes language models, multimodal models and biological language models.
-
As of October 2026
Languages and computing
Belinkov studied mathematics and Arabic at Tel Aviv University as an undergraduate, then completed a master’s thesis on an Arabic dialect. He also worked as an IntuView software engineer.
-
As of October 2026
Doctoral and postdoctoral research
Belinkov completed his MIT doctorate under James Glass. His thesis examined internal language representations in deep-learning models. He later worked as a Harvard postdoctoral researcher with Stuart Shieber.
Sources and credits
Photo: ellis.eu; Yonatan Belinkov profile photograph; creator not stated. License: Not stated by source. Source: ellis.eu. Cropped to a square and resized. Ask us to remove or replace this photo.