Lukasz Kaiser

Kaiser was born in Wrocław, Poland. He studied computer science and mathematics at the University of Wrocław, receiving master's degrees in both subjects in 2003.^[2] His mathematics thesis was supervised by Grzegorz Plebanek, and his computer science thesis by Leszek Pacholski.^[5]

From 2003 to 2008, he pursued doctoral studies at RWTH Aachen University under Erich Grädel.^[6] His dissertation, Logic and Games on Automatic Structures, studied games and logical definability over automatic structures.^[7] It received the 2009 E. W. Beth Dissertation Prize.^[8]

After receiving his doctorate, Kaiser held postdoctoral and research-project positions at RWTH Aachen connected with algorithmic strategies, synthesis, and automata-theoretic games.^[9] In October 2010, he joined CNRS as a chargé de recherche at LIAFA, Université Paris Diderot, where he held a tenured research position before moving into machine learning.^[2]

In 2013, Kaiser joined Google Research to work under Ray Kurzweil and later moved to Google Brain to work under Ilya Sutskever. By 2018, he was a staff research scientist on the Google Brain team.^[10] He was among the authors of the TensorFlow system paper and Google's Neural Machine Translation system paper.^[11]

Kaiser joined OpenAI in June 2021.^[12] He was acknowledged among the contributors to ChatGPT at its launch on 30 November 2022.^[9][13]

In 2017, Kaiser co-authored "Attention Is All You Need" with Ashish Vaswani, Noam Shazeer, Niki Parmar, Jakob Uszkoreit, Llion Jones, Aidan N. Gomez, and Illia Polosukhin.^[14] The paper introduced the Transformer, a neural-network architecture based on attention mechanisms rather than recurrence or convolution.^[3]

At OpenAI, Kaiser was one of the leading researchers who developed OpenAI o1, which has been described as the first publicly released reasoning model.^[15] He later contributed to o3, o4-mini, and GPT-5.^[16]

• 2009 – E. W. Beth Dissertation Prize for Logic and Games on Automatic Structures.^[8]
• 2024 – NEC C&C Prize, as a member of the Transformer Team, for pioneering research on the Transformer deep-learning model.^[4]

Logic, automata theory, and games

• Kaiser, Łukasz (2011). Logic and Games on Automatic Structures: Playing with Quantifiers and Decompositions. Lecture Notes in Computer Science. Vol. 6810. Springer. doi:10.1007/978-3-642-22774-1 (inactive 19 July 2026).{{cite book}}: CS1 maint: DOI inactive as of July 2026 (link)
• Kaiser, Łukasz. Game Quantification on Automatic Structures and Hierarchical Model Checking Games. CSL 2006.
• Kaiser, Łukasz; Rubin, Sasha; Bárány, Vince. Cardinality and Counting Quantifiers on Omega-Automatic Structures. STACS 2008.
• Fischer, Diana; Grädel, Erich; Kaiser, Łukasz (2010). "Model Checking Games for the Quantitative μ-Calculus". Theory of Computing Systems. 47 (3): 696–719. doi:10.1007/s00224-009-9201-y.
• Berwanger, Dietmar; Kaiser, Łukasz (2010). "Information Tracking in Games on Graphs". Journal of Logic, Language and Information. 19 (4): 395–412. doi:10.1007/s10849-009-9115-8.
• Abu Zaid, Faried; Grädel, Erich; Kaiser, Łukasz. The Field of Reals is not Omega-Automatic. STACS 2012.

Machine learning and Transformers

• Kaiser, Łukasz; Sutskever, Ilya. Neural GPUs Learn Algorithms. ICLR 2016.
• Kaiser, Łukasz; Bengio, Samy (2016). Can Active Memory Replace Attention?. Advances in Neural Information Processing Systems.
• Kaiser, Łukasz; Nachum, Ofir; Roy, Aurko; Bengio, Samy. Learning to Remember Rare Events. ICLR 2017.
• Vaswani, Ashish; Shazeer, Noam; Parmar, Niki; Uszkoreit, Jakob; Jones, Llion; Gomez, Aidan N.; Kaiser, Łukasz; Polosukhin, Illia (2017). Attention Is All You Need. Advances in Neural Information Processing Systems.
• Kitaev, Nikita; Kaiser, Łukasz; Levskaya, Anselm. Reformer: The Efficient Transformer. ICLR 2020.