Ioannis Antonoglou
Ioannis Antonoglou is co-founder and CTO of Reflection AI, which develops open foundation models. He contributed to DeepMind papers on game-playing systems, including the Deep Q-Network, AlphaGo Zero, and MuZero. After studying AI at the University of Edinburgh, he joined DeepMind in 2012 as employee number 25. In 2024 he co-founded Reflection AI with Misha Laskin, and the company raised $2 billion at an $8 billion valuation in October 2025. At DeepMind he also led the team responsible for reinforcement learning from human feedback on Gemini.
Timeline
-
2012
Joining DeepMind early
He completes his AI studies at the University of Edinburgh and joins DeepMind in 2012 as employee number 25.
-
2015
The Deep Q-Network paper
He co-authors the deep Q-network paper, which Nature publishes in February 2015. The agent learns to play Atari games directly from pixels.
-
2017
AlphaGo Zero learns from scratch
He co-authors AlphaGo Zero, which learns Go by playing itself from random play. Nature publishes the paper in October 2017.
-
2020
MuZero plans with a learned model
He co-authors MuZero, a Nature paper published in December 2020 on planning with a learned model.
-
2023–2024
Reinforcement learning for Gemini
He leads the team responsible for reinforcement learning from human feedback on Gemini. Google has announced Gemini by December 2023, and v1.0 is released in February 2024.
-
2024
Co-founding Reflection AI
He co-founds Reflection AI with Misha Laskin in 2024.
-
October 2025
A $2 billion funding round
Reflection AI raises $2 billion at an $8 billion valuation in October 2025.
-
As of October 2026
Co-founder and CTO
He is co-founder and CTO of Reflection AI, which develops open foundation models and software agents for AI-assisted software development.
Sources and credits
Photo: We have no published photo of Ioannis Antonoglou yet. Send us a free-licensed photo.
- Ioannis Antonoglou | Entrepreneur | | Endeavor Greece
- Reflection AI: The Race to Unlock Superintelligence | Sequoia Capital
- Human-level control through deep reinforcement learning | Nature
- AlphaGo Zero: Starting from scratch — Google DeepMind
- Mastering the game of Go without human knowledge | Nature
- Mastering Atari, Go, chess and shogi by planning with a learned model | Nature
- Reflection AI - Wikipedia
- Inside Reflection AI: The $20B Open-Model Startup That Has Yet to Ship