{"version":1,"id":"jakucb-pachoki","title":"Jakucb Pachoki","summary":"Jakub Pachocki is a computer scientist and OpenAI chief scientist known for leading AI research and contributing to the development of models including GPT-4.","content":"**Jakub Pachocki** is a Polish computer scientist and former competitive programmer who is chief scientist at OpenAI. Born in Gdańsk in 1991, he joined the company in 2017 and has held research leadership roles spanning deep learning, reasoning, and model development.[\\[1\\]](#cite-id-a1vjokyqdv)​[\\[2\\]](#cite-id-gyr0kbenfh)​\n\n## Early Life and Education\n\nPachocki was born in Gdańsk, Poland, in 1991. During secondary school, he was a six-time finalist in Poland’s Informatics Olympiad, where Ryszard Szubartowski was his teacher. He also won programming competitions, including Google Code Jam in 2012.[\\[2\\]](#cite-id-gyr0kbenfh)​\n\nHe studied computer science at the University of Warsaw from 2010 to 2013, then pursued a doctorate at Carnegie Mellon University (CMU) from 2013 to 2016. He earned his Ph.D. in May 2016 with a dissertation titled _Graphs and Beyond: Faster Algorithms for High Dimensional Convex Optimization_. Gary Miller advised the dissertation and chaired its committee; the other committee members included Anupam Gupta, Daniel Sleator, Shang-Hua Teng, Frank Pfenning, and Andrew W. Moore.[\\[1\\]](#cite-id-a1vjokyqdv)​[\\[3\\]](#cite-id-txyxvtm9oy)​\n\nThe dissertation addressed algorithm design for large and complex data. It developed nearly optimal algorithms for the Fermat–Weber, or geometric median, problem; established approximation tradeoffs for high-dimensional data in restrictive streaming settings; and presented a method for solving linear systems on undirected graphs. It also established clustering and oblivious-routing algorithms for a broad family of directed graphs, with potential relevance to faster maximum-flow algorithms. Most of its algorithms ran in time nearly linear in the input’s sparsity, and its methods combined combinatorial and numerical analysis.[\\[3\\]](#cite-id-txyxvtm9oy)​\n\nPachocki was a postdoctoral fellow at Harvard University from July 2016 to January 2017. He also had an association with the Simons Institute for the Theory of Computing.[\\[1\\]](#cite-id-a1vjokyqdv)​[\\[2\\]](#cite-id-gyr0kbenfh)​\n\n## Research and OpenAI Career\n\nBefore joining OpenAI, Pachocki worked as a software engineer at Facebook from 2011 to 2012, according to a 2025 profile.[\\[2\\]](#cite-id-gyr0kbenfh) He joined OpenAI in February 2017 as research lead for its Dota team, a position he held until March 2019. He then led the reasoning team from March 2019 to March 2021 and served as research lead for the Science of Deep Learning from March 2021 to July 2022.[\\[1\\]](#cite-id-a1vjokyqdv)​\n\nPachocki became a principal of research in November 2021 and held the role until October 2023. He was director of research from October 2023 to May 2024, when he became OpenAI’s chief scientist.[\\[1\\]](#cite-id-a1vjokyqdv) A 2025 report also describes him as having led development of GPT-4 and OpenAI Five and as having made a significant contribution to ChatGPT; his professional profile identifies his role overseeing GPT-4 development.[\\[2\\]](#cite-id-gyr0kbenfh)​[\\[1\\]](#cite-id-a1vjokyqdv)​\n\nPachocki worked closely with OpenAI co-founder [Ilya Sutskever](https://iq.wiki/wiki/ilya-sutskever) on research projects, including an advanced-reasoning model known as Q\\*, later renamed Strawberry. He succeeded Sutskever as chief scientist after Sutskever’s departure from OpenAI amid controversy over [Sam Altman](https://iq.wiki/wiki/sam-altman)’s removal and subsequent return as chief executive. Pachocki described Sutskever as his mentor and collaborator, crediting him with introducing him to deep-learning research.[\\[4\\]](#cite-id-dg9e8y5ti4)​[\\[2\\]](#cite-id-gyr0kbenfh)​\n\nPachocki’s research has included artificial intelligence, computer vision, algorithms, graph theory, computational mathematics, data compression, and automata theory. His publications range from theoretical computer science to large-scale machine learning. Early work included “A Note on Efficient Computation of All Abelian Periods in a String” (2012), “Online Row Sampling” (2016), and “Optimal Lower Bounds for Universal Relation, and for Samplers and Finding Duplicates in Streams” (2017).[\\[1\\]](#cite-id-a1vjokyqdv)​\n\nHis AI publications include “_Emergent Complexity via Multi-Agent Competition” (2017), “Learning Dexterous In-Hand Manipulation_” (2019), “_Dota 2 with Large Scale Deep Reinforcement Learning_” (2019), and “_Tensor Programs V: Tuning Large Neural Networks via Zero-Shot Hyperparameter Transfer”_ (2022). Later works include the GPT-4o System Card and OpenAI o1 System Card, both published in 2024, and model cards for gpt-oss-120b and gpt-oss-20b and the OpenAI GPT-5 System Card, published in 2025.[\\[1\\]](#cite-id-a1vjokyqdv)​\n\n## Recognition\n\nPachocki was included in Business Insider’s 2024 AI Power List. In 2023, he was named among 23 influential Polish figures in AI by _My Company Polska_.[\\[4\\]](#cite-id-dg9e8y5ti4)​[\\[2\\]](#cite-id-gyr0kbenfh) On September 9, 2026, he welcomed Paul Christiano’s appointment to the OpenAI Foundation board and said the organization had substantial work ahead.[\\[6\\]](#cite-id-kyo5dvz26x)​\n","categories":[{"id":"people","title":"people"}],"tags":[{"id":"Developer"},{"id":"Researcher"}],"images":[{"id":"QmeTfCRrYYj2udjHqKKph7nGJ5dHk6XrBJMbpcJuoJko5X","type":"image/jpeg, image/png"}],"media":[],"metadata":[{"id":"references","value":"[{\"id\":\"a1vjokyqdv\",\"description\":\"LinkedIn: Jakub Pachoki\",\"timestamp\":1792822726455,\"url\":\"https://www.linkedin.com/in/jakub-pachocki/\"},{\"id\":\"gyr0kbenfh\",\"description\":\"Wirtual Media - Jakub Pachocki - Chief Scientist w OpenAI\",\"timestamp\":1792822747589,\"url\":\"https://www.wirtualnemedia.pl/jakub-pachocki-chief-scientist-w-openai,7176128594118785a\"},{\"id\":\"txyxvtm9oy\",\"description\":\"CMU - Graphs and Beyond: Faster Algorithms for High Dimensional Convex Optimization\",\"timestamp\":1792822766638,\"url\":\"https://csd.cmu.edu/academics/doctoral/degrees-conferred/jakub-pachocki\"},{\"id\":\"dg9e8y5ti4\",\"description\":\"Business Insider - Jakub Pachocki, Chief Scientist at OpenAI\",\"timestamp\":1792822783606,\"url\":\"https://www.businessinsider.com/jakub-pachocki-openai-ai-power-list-2024\"},{\"id\":\"itmnirnyhw\",\"description\":\"Jakub Pachoki - “An Alien Mind” \",\"timestamp\":1792822801522,\"url\":\"https://openai.com/index/an-alien-mind/\"},{\"id\":\"kyo5dvz26x\",\"description\":\"Jakub Pachocki X Account\",\"timestamp\":1792822810271,\"url\":\"https://x.com/merettm\"}]"},{"id":"main-image-origin","value":"{\"imageId\":\"QmeTfCRrYYj2udjHqKKph7nGJ5dHk6XrBJMbpcJuoJko5X\",\"originalId\":\"QmeTfCRrYYj2udjHqKKph7nGJ5dHk6XrBJMbpcJuoJko5X\"}"},{"id":"twitter_profile","value":"https://x.com/merettm"},{"id":"linkedin_profile","value":"https://www.linkedin.com/in/jakub-pachocki/"},{"id":"commit-message","value":"Add the Jakucb Pachoki wiki"}],"events":[{"type":"DEFAULT","title":"Became OpenAI chief scientist","date":"2024-05-01","description":"Pachocki became chief scientist at OpenAI.","link":null,"multiDateStart":null,"multiDateEnd":null,"id":"40d0f14e-b5b5-4c9d-82b4-deb83f1cec28"}],"linkedWikis":{"founders":[],"blockchains":[],"speakers":[]},"user":{"id":"0x8af7a19a26d8fbc48defb35aefb15ec8c407f889"},"author":{"id":"0x4a2bc97afbf41a2cbf4b4628f63420682ff4cf2a"},"language":"en","operator":{"id":"0x4a2bc97afbf41a2cbf4b4628f63420682ff4cf2a"}}