ABAB Pedia

Misha Laskin

Overview

Contents11 sections
Key facts

Early life: From Russia to Israel and the United States

Education and early ventures: Physics, transportation, and machine learning

Career: From DeepMind to Reflection AI

At Google DeepMind he was a research scientist on Gemini, studying how reinforcement learning could expand the abilities of language and multimodal models. In 2024 he and Ioannis AntonoglouIoannis AntonoglouIoannis Antonoglou (full name Ioannis Alexandros Antonoglou; Greek: Ιωάννης Αντώνογλου) is an AI researcher from Thessaloniki. He helped develop the Go program AlphaGo at DeepMind and co-founded Reflection AI with Misha Laskin.Open the full entry founded Reflection AI to make models available for developers and institutions to use and adapt. Its releases included Asimov for understanding codebases, followed by the open-weight reasoning model Beam.⁠[1][2][5]

Interview: No Priors

Interview: No Priors

He returned on October 9, 2026, to discuss Beam, reasoning efficiency, enterprise use of open-weight models, and competition between open and closed systems.⁠[5]

Advice Shared on No Priors

He advised designing evaluations around what customers actually ask and do, such as the questions engineers raise while joining a company. (17:57)⁠[6]

He argued that a research team needs a shared mission and culture to retain talented people and keep them working together. (03:11)⁠[5]

Related Organizations, Websites and Public Accounts

Reflection AI: Co-founder and chief executive. Official website: https://reflection.ai/ (opens in a new window).⁠[2]

His personal website is [mishalaskin.com](https://www.mishalaskin.com/ (opens in a new window)).⁠[1]

His public X profile is [@MishaLaskin](https://x.com/MishaLaskin (opens in a new window)).⁠[1]

Interview with No Priors

He said Reflection worked backward from how a product would be used and designed the research alongside it. This connected the team to real problems early but narrowed its initial scope.⁠[6]

Interview with No Priors

He said a lab could not scale simply by hiring. The compute available per researcher mattered, and thousands of hires would not necessarily be more useful than a few hundred.⁠[5]

Sources7 sources

  1. Misha Laskin: About (opens in a new window)mishalaskin.com↑
  2. Reflection AI | Sequoia Capital (opens in a new window)sequoiacap.com↑
  3. Ex-DeepMind Researcher Misha Laskin on Enterprise Super-Intelligence (opens in a new window)mattturck.com↑
  4. VanGo: Uber for Kids (opens in a new window)ycombinator.com↑
  5. Beam: The Great American Open Model with ReflectionAI Co-Founder and CEO Misha Laskin (opens in a new window)youtube.com↑
  6. No Priors Ep. 123 | With ReflectionAI Co-Founder and CEO Misha Laskin (opens in a new window)youtube.com↑
  7. Misha Laskin — ReflectionAI (opens in a new window)linkedin.com↑