PeopleofSuperintelligence

Researcher

Konstantin Mishchenko

Konstantin Mishchenko is an optimization researcher whose work on learning-rate-free methods lets neural networks train without hand-tuned step sizes. He earned a BSc from the Moscow Institute of Physics and Technology in 2016, a master's degree in Paris in 2017, and a PhD at KAUST in 2021 under Peter Richtárik. After a postdoc at Inria Sierra and research at the Samsung AI Center in Cambridge, he won an ICML Outstanding Paper Award in 2023 with Aaron Defazio for D-Adaptation, then followed it with Prodigy in 2023 and Schedule-Free learning in 2024. He is a research scientist at Meta in Paris, working on code generation.

Now at
Meta
Badges

Timeline

  1. 2016–2017

    Degrees in Moscow and Paris

    He earns a BSc in computer science and physics from the Moscow Institute of Physics and Technology in 2016. In 2017, he completes a double-degree MSc in data science at École normale supérieure Paris-Saclay and Paris-Dauphine.

  2. 2017–2021

    A PhD at KAUST

    He works at KAUST under Peter Richtárik on optimization theory and its applications in machine learning, and he completes his computer science PhD in 2021 with a thesis titled On Seven Fundamental Optimization Challenges in Machine Learning. In 2020, he interns at Google Brain.

  3. December 2021–December 2022

    Postdoc at Inria Sierra

    He is a postdoc at Inria Sierra in Paris with Alexandre d'Aspremont and Francis Bach, researching adaptive, second-order, and distributed optimization.

  4. January 2023–October 2024

    Research scientist at Samsung

    At the Samsung AI Center in Cambridge, UK, he works on embedded AI systems as a member of the Distributed AI team and the GenAI initiative. His adaptive optimization methods for automated training appear at ICML 2023 and ICML 2024.

  5. 2023

    ICML Outstanding Paper Award

    Learning-Rate-Free Learning by D-Adaptation, written with Aaron Defazio, receives an Outstanding Paper award at ICML 2023. The method matches hand-tuned learning rates across more than a dozen machine learning problems without knowing the distance from the starting point to the solution.

  6. June 2023

    The Prodigy optimizer

    In June 2023, he and Aaron Defazio post Prodigy, a modification of D-Adaptation that improves its convergence rate and estimates the learning rate for methods such as AdaGrad and Adam.

  7. May 2024

    The Road Less Scheduled

    In May 2024, he coauthors The Road Less Scheduled with Aaron Defazio, Xingyu Alice Yang, Harsh Mehta, Ahmed Khaled and Ashok Cutkosky, which introduces Schedule-Free learning without learning-rate schedules. Schedule-Free AdamW is the core algorithm behind the team's winning entry to the MLCommons 2024 AlgoPerf Self-Tuning track.

  8. Since October 2024

    Research scientist at Meta

    He joins Meta in Paris in October 2024 as a research scientist, doing research on code generation.

Career

  1. Since 2024MetaResearch scientist
  2. 2023–2024Samsung
  3. 2021–2022InriaPostdoctoral researcher
  4. 2020Google BrainIntern

Education

  1. 2021King Abdullah University of Science and TechnologyPhD
  2. 2017École Normale Supérieure Paris-SaclayMSc
  3. 2017Paris Dauphine UniversityMSc
  4. 2016Moscow Institute of Physics and TechnologyBSc

From the biography.

  1. Of 943 people shaping AI who worked at Google, 446 work there nowIncludes Konstantin Mishchenko. Our estimates of where the 943 people shaping AI who worked at Google and Google DeepMind work now: 446 at Google and 452 across 278 other organizations.
  2. Google alumni have founded or led 133 other organizations, more than any other AI labIncludes Konstantin Mishchenko. We estimate that alumni of 17 major AI labs went on to found or lead 350 other organizations, including 226 they founded, and we trace 472 career moves directly between 23 labs since 2010.
  3. 16% of AI career moves in the 2020s went from a university to a companyIncludes Konstantin Mishchenko. We estimate that 16% of AI career moves in the 2020s went from a university to a company; 29% of moves starting at a university went to a company, and 76% ended the university role.
  4. The Massachusetts Institute of Technology trained or employed 44 of the 187 AI major-prize winners, more than any other universityIncludes Konstantin Mishchenko. The Massachusetts Institute of Technology trained or employed 44 of the 187 AI major-prize winners, more than any other university. We compare doctorate and 5+ year employment links, career timing and current workplaces for 187 major-prize and 1,111 best-paper winners.

Colleagues

People named in this biography, and people with dated positions at the same organization in the same years.

Sources and credits

Photo: We have no published photo of Konstantin Mishchenko yet. Send us a free-licensed photo.

  1. Konstantin Mishchenko konstmish.com. Checked 21 October 2026.
  2. Peter Richtárik — Team richtarik.org. Checked 21 October 2026.
  3. ICML 2023 Awards icml.cc. Checked 21 October 2026.
  4. [2301.07733] Learning-Rate-Free Learning by D-Adaptation arxiv.org. Checked 21 October 2026.
  5. [2306.06101] Prodigy: An Expeditiously Adaptive Parameter-Free Learner arxiv.org. Checked 21 October 2026.
  6. [2405.15682] The Road Less Scheduled arxiv.org. Checked 21 October 2026.