Jakub Pachocki

Jakub Pachocki is a Polish computer scientist, born in 1991 in Gdańsk, Poland, who serves as Chief Scientist of OpenAI. He led the development of GPT-4 and OpenAI Five as Director of Research, and has been one of the central architects of the o-series reasoning-model program since o1-preview. As of May 2026, he leads OpenAI's research direction alongside Chief Research Officer Mark Chen, having succeeded Ilya Sutskever in May 2024, and is named on the TIME 100 AI list for 2025.

At a glance

• Education: Master's degree in computer science, University of Warsaw (2013); PhD in computer science, Carnegie Mellon University (2016), advised by Gary L. Miller, with the thesis Graphs and Beyond: Faster Algorithms for High Dimensional Convex Optimization.
• Current role: Chief Scientist of OpenAI since May 2024, having previously been Director of Research from 2021.
• Key contributions: silver medal at the 2009 International Olympiad in Informatics; gold medal and second place overall at the 2012 ACM-ICPC World Finals with the University of Warsaw team; champion of Google Code Jam 2012; lead author of OpenAI Five; overall lead and optimization lead for GPT-4; research direction for the o-series reasoning models.
• X / Twitter: @merettm
• Personal site: scholar.harvard.edu/meret
• LinkedIn: Jakub Pachocki
• Wikipedia: Jakub Pachocki

Origins

Pachocki was born in 1991 in Gdańsk, on Poland's Baltic coast, and entered Polish national informatics competitions in school. He reached the finals of the International Olympiad in Informatics multiple times, taking a silver medal in the 2009 edition in Plovdiv, Bulgaria. The competitive-programming background remains a defining feature of his public profile and of OpenAI's reasoning-model research culture, where many senior staff hold International Olympiad in Informatics or International Mathematical Olympiad records.

He completed a master's degree in computer science at the University of Warsaw in 2013. The Warsaw team of Pachocki, Tomasz Kulczyński, and Wojciech Śmietanka took the gold-medal first-place finish by problems solved at the 2012 ACM-ICPC World Finals in Warsaw, second overall on time penalty behind St. Petersburg State University.

Career

Pachocki entered Carnegie Mellon University's computer science PhD program in the mid-2010s, working with Gary L. Miller. His doctoral research focused on faster algorithms for high-dimensional convex optimization and graph problems, with publications on Laplacian solvers, maximum-flow algorithms, and spectral graph theory at STOC, FOCS, and SODA. He defended his thesis, Graphs and Beyond: Faster Algorithms for High Dimensional Convex Optimization, in 2016.

A brief postdoctoral period followed at Harvard University and at the Simons Institute for the Theory of Computing at UC Berkeley, where his research interests turned toward algorithms for machine-learning applications. Earlier industry exposure included a software-engineering internship at Facebook between 2011 and 2012.

In 2017 Pachocki joined OpenAI as a research scientist, moving from theoretical computer science into large-scale machine-learning research. His earliest work focused on reinforcement learning at scale, and he became the technical lead on OpenAI Five, the Dota 2 reinforcement-learning system that defeated the OG world-champion human team in April 2019. The 2019 paper, Dota 2 with Large Scale Deep Reinforcement Learning, lists Pachocki among the lead authors and was widely cited as an early demonstration that engineering-scale reinforcement learning could produce superhuman performance in complex multi-agent environments.

In 2021 he was promoted to Director of Research at OpenAI, taking responsibility for the deep-learning research direction across language modeling, reinforcement learning, and optimization. OpenAI's GPT-4 contributors page credits Pachocki as overall lead and optimization lead for the multimodal flagship released in March 2023. By mid-2023 he was driving the company's pivot toward reasoning models, the line that produced o1-preview (September 2024) and o3 (early 2025) and that has remained the dominant frontier-research thread at the lab through GPT-5 and GPT-5.5.

On May 14, 2024, OpenAI announced that Sutskever was departing the company and that Pachocki would succeed him as Chief Scientist. CEO Sam Altman wrote in the announcement that Pachocki was "easily one of the greatest minds of [his] generation," with central contributions to GPT-4, OpenAI Five, deep-learning optimization, and the reasoning research line. The transition placed Pachocki at the head of OpenAI's long-term research roadmap during the Superalignment team's disbandment in the same month and through the senior-staff departures of Mira Murati, John Schulman, Bob McGrew, Lilian Weng, Barret Zoph, and Liam Fedus across the rest of 2024 and into 2025.

Affiliations

• Facebook: Software engineering intern, 2011 to 2012.
• Harvard University: Postdoctoral researcher, 2016 to 2017 (approximate).
• Simons Institute for the Theory of Computing: Visiting researcher, 2017.
• OpenAI: Research Scientist, 2017 to 2021.
• OpenAI: Director of Research, 2021 to 2024-05.
• OpenAI: Chief Scientist, 2024-05 to present.

Notable contributions

Pachocki's body of public work spans algorithmic theory, large-scale reinforcement learning, language-model pretraining, and reasoning-model research at one of the central frontier AI labs.

• Theoretical computer science (2013 to 2017). Co-author of papers on faster Laplacian solvers, maximum-flow algorithms, and approximate matching at STOC, FOCS, and SODA, with collaborators including Gary L. Miller, Yin Tat Lee, Jonathan Kelner, and Aaron Sidford.
• Competitive-programming record. Silver medal at the 2009 International Olympiad in Informatics in Plovdiv; gold medal and second-place overall at the 2012 ACM-ICPC World Finals with the University of Warsaw team; champion of Google Code Jam 2012 and third place in 2011; second place at the Facebook Hacker Cup 2013 and the TopCoder Open Algorithm 2012. Online handle "meret" on TopCoder, Codeforces, and CodeChef.
• OpenAI Five (2017 to 2019). Lead author with Christopher Berner, Greg Brockman, Brooke Chan, and others on the Dota 2 reinforcement-learning system. The April 2019 result against the OG world-champion team was an early demonstration of large-scale self-play training.
• GPT-4 (March 2023). Overall research lead and optimization lead for the multimodal flagship, per the GPT-4 contributors page. The model defined the post-ChatGPT capability frontier through 2023 and 2024.
• o-series reasoning research (2023 to present). Senior research direction for the line of models that introduced reinforcement learning on chain-of-thought reasoning. The September 2024 o1-preview release and the subsequent o3 and o4 models established a new capability axis distinct from pretraining-scale increases. In a July 2025 post, Pachocki cited chain-of-thought faithfulness and interpretability as a central design principle for the line.
• Competitive-programming and mathematics benchmarks (2025). OpenAI's reasoning models took gold medals at the 2025 International Mathematical Olympiad and the International Olympiad in Informatics, and a perfect 12-of-12 score at the 2025 ACM-ICPC World Finals in Baku, a result that would have placed first in the world.
• TIME 100 AI (2025). Named to the TIME 100 AI list of the most influential people in artificial intelligence.

Investments and boards

No public personal investor activity on record in AI, semiconductors, datacenters, software, or energy as of May 2026. Pachocki's footprint in this section is concentrated in his research and operating role at OpenAI rather than a parallel investing program.

Network

Pachocki's foundational professional relationship is with his Carnegie Mellon PhD advisor Gary L. Miller, with whom he co-authored multiple papers in the algorithmic graph-theory and convex-optimization line. The CMU and broader theoretical-computer-science circle, including Yin Tat Lee, Jonathan Kelner, and Aaron Sidford, is the academic peer group from his pre-OpenAI period.

His OpenAI peer cohort, with whom he has worked since 2017, includes Sam Altman, the chief executive who appointed him; Greg Brockman, the president and OpenAI Five engineering lead; Ilya Sutskever, his predecessor as Chief Scientist and research mentor, who departed in May 2024 to found Safe Superintelligence; Mark Chen, the Chief Research Officer with whom he shares OpenAI's twin research-leadership structure; and Wojciech Zaremba, the Polish-born OpenAI co-founder and reinforcement-learning researcher. The 2024 senior-departure cohort included Mira Murati (now Thinking Machines Lab), John Schulman, Lilian Weng, Barret Zoph, and Andrej Karpathy, with Liam Fedus departing in March 2025.

The competitive-programming community is a separate professional network. Pachocki is identified in Polish technology coverage alongside Przemysław Dębiak ("Psyho"), Marek Cygan, and other Poland-origin algorithmic-competition figures, several of whom have followed similar paths into AI research at OpenAI, Google DeepMind, and other frontier labs.

Position in the field

As of May 2026, Pachocki is one of two named research leaders at OpenAI, sharing the senior-research function with Chief Research Officer Mark Chen since September 2024. The MIT Technology Review July 2025 profile of the pair characterized the division of labor as Pachocki setting long-term technical direction and roadmap while Chen runs research operations and manages immediate company needs. The pairing is structurally distinct from the Sutskever-era single-Chief-Scientist model that prevailed through May 2024.

His public profile concentrates on research-direction work rather than the policy-and-media circuit. Industry coverage including the May 2025 Nature interview on automated scientific research, the July 2025 MIT Technology Review profile, and the September 2025 ICPC announcement on the perfect-score reasoning-model result frames him as the senior technical voice for OpenAI's research roadmap. Sam Altman's May 2024 blog post co-named Pachocki and OpenAI engineer Szymon Sidor as central to OpenAI Five, GPT-4 pretraining, and the reasoning breakthroughs.

The competitive-programming heritage is unusual among Frontier-lab research leaders. The 2025 ICPC perfect-score result and the matching International Mathematical Olympiad and International Olympiad in Informatics gold medals are widely cited as evidence that the reasoning-model research program Pachocki leads has reached competitive-mathematics performance previously thought to require dedicated specialist systems.

Outlook

Open questions over the next 6 to 18 months:

• Reasoning-model trajectory. Whether the o-series cadence produces continued capability gains on competitive-mathematics, scientific-research, and software-engineering benchmarks beyond the 2025 results, and whether the line converges with the GPT pretraining-scale family.
• Automated-research roadmap. OpenAI's stated goal of an "AI researcher" capable of independent scientific work has been described as the company's research north star. Public artifacts against the 2026-to-2028 timeline are watchable signals.
• Senior-research stability. Whether the twin-leadership model with Mark Chen continues unchanged, or whether further senior departures require restructuring the research function.
• Public commentary frequency. Whether the MIT Technology Review profile, the Nature interview, and the TIME 100 AI 2025 recognition mark a sustained shift in public visibility from Pachocki's previously low-profile cadence.
• Safety and alignment posture. The future of OpenAI's alignment research after the May 2024 Superalignment disbandment, and Pachocki's direction on the chain-of-thought-faithfulness and interpretability program he has cited for the o-series.
• Long-term role at OpenAI. Whether Pachocki remains Chief Scientist through the GPT-6 cycle and the corporate restructuring that has continued through 2025 and 2026.

Sources

• Jakub Pachocki. Wikipedia biographical entry covering education, the competitive-programming record, the OpenAI roles, and the Chief Scientist appointment.
• Ilya Sutskever to leave OpenAI, Jakub Pachocki announced as Chief Scientist. OpenAI's May 14, 2024 announcement of the leadership transition with Sam Altman's framing.
• Jakub and Szymon. Sam Altman's May 2024 blog post recognizing Pachocki and Szymon Sidor for their work on OpenAI Five, GPT-4 pretraining, and the reasoning breakthroughs.
• Graphs and Beyond: Faster Algorithms for High Dimensional Convex Optimization. Pachocki's 2016 Carnegie Mellon PhD thesis under Gary L. Miller.
• Dota 2 with Large Scale Deep Reinforcement Learning. The 2019 OpenAI Five paper with Pachocki among the lead authors.
• GPT-4 contributions. OpenAI's GPT-4 contributors page listing Pachocki as overall lead and optimization lead.
• Learning to Reason with LLMs. OpenAI's September 2024 announcement of o1-preview and the reasoning-model line.
• The two people shaping the future of OpenAI's research. MIT Technology Review July 2025 profile of Mark Chen and Jakub Pachocki.
• AI models are capable of novel research: OpenAI's chief scientist on what to expect. Nature interview, May 2025, on automated scientific research.
• Jakub Pachocki: The 100 Most Influential People in AI 2025. TIME 100 AI 2025 entry on Pachocki as Chief Scientist of OpenAI.
• Jakub Pachocki on the 2025 ICPC World Finals result. The September 2025 X post on OpenAI's reasoning-model perfect-score result at the ICPC.
• Jakub Pachocki on chain-of-thought faithfulness and interpretability. The July 2025 X post on the design principles behind the o-series.
• From Vibe Coding to Vibe Researching: OpenAI's Mark Chen and Jakub Pachocki. September 2025 conversation with Mark Chen and Jakub Pachocki on OpenAI's research direction.
• Jakub Pachocki at the Simons Institute for the Theory of Computing. Simons Institute profile from Pachocki's 2017 visiting-researcher period.