معرفی
Samet Oymak is an Associate Professor of Electrical Engineering and Computer Science at the University of Michigan, Ann Arbor (effective August 2025). Previously, he was an Assistant Professor of Electrical and Computer Engineering at the University of California, Riverside. His research focuses on the intersection of machine learning, optimization, and statistics, with a recent emphasis on trustworthy AI, transformers, and reinforcement learning.
Education:
- B.S. in Electrical Engineering, Bilkent University (2009)
- M.S. and Ph.D. in Electrical Engineering, California Institute of Technology (2011 and 2015)
Research Interests: Oymak’s work addresses foundational challenges in AI/ML, including transformer theory, sequence modeling, and the capabilities of language models. He explores optimization and statistical learning theory to enhance model efficiency, robustness, and reasoning capabilities. Current projects include: parallel exploration in reasoning, bridging SFT & RL for reasoning efficiency, and improving in-context learning through unlabeled data.
Recent Trends in Articles: His recent work emphasizes provable guarantees for in-context learning, attention mechanisms, and adaptive multi-source tuning. Key themes include transformer dynamics, reinforcement learning for small models, and trustworthiness in environmental AI applications.
Awards:
- EECS Outstanding Achievement Award (2025)
- Amazon Research Award (2024)
- Adobe Data Science Research Award (2023)
- NSF CAREER Award (2021)
- Simons-Berkeley Research Fellowship (2015)
Advising & Grants: Oymak has advised students like Mingchen (now at Meta) and secured NSF and ONR grants for foundational LLM research. His lab focuses on theoretical and algorithmic advancements in AI.
Labs & Teams: His SOTA Lab develops principled methods for trustworthy AI, with a focus on transformers and time-series forecasting. Recent collaborations include work on Pyneapple-L for spatial analysis and TimePFN for multivariate forecasting.
