From AlphaGo to Autonomous Agents: The Journey of Ioannis Antonoglou

Ioannis Antonoglou is one of the most influential yet understated forces in modern artificial intelligence. Born and raised in Thessaloniki, he built the foundation of his scientific path at the Aristotle University of Thessaloniki before moving to the University of Edinburgh, where he completed his studies in Artificial Intelligence. What followed was a leap that would shape the next decade of global AI research.

In late 2012, Ioannis joined DeepMind as one of its earliest researchers - employee number 25 and just the sixth member of the research team. At a time when the company was still a bold experiment in London, he chose the uncertainty of frontier research over safer academic paths. That choice placed him at the epicenter of some of the most historic breakthroughs in AI.

During his twelve years at DeepMind, Ioannis became a core contributor to milestone achievements that defined a new era for the field: the Deep Q-Network (DQN) that mastered Atari games directly from pixels; AlphaGo, the system that defeated world champion Lee Sedol; AlphaGo Zero and AlphaZero, which learned Go, chess, and shogi from scratch without human examples; and MuZero, a model that learned to plan and play games without even knowing the rules. Later, when Google DeepMind launched Gemini, Ioannis led the team responsible for Reinforcement Learning from Human Feedback (RLHF), helping align Google’s frontier language models with human intentions.

After more than a decade at the world’s leading AI lab, Ioannis took his biggest professional leap yet. In 2024, together with fellow DeepMind researcher Misha Laskin, he co-founded Reflection AI, headquartered in New York. Their mission: to build superintelligent autonomous systems, beginning with the most technically demanding domain - software engineering.

Reflection AI’s first system, Asimov, represents a new class of AI agent. Instead of merely generating code, Asimov ingests entire codebases, documentation, design specs, and team conversations to understand how complex software systems work. Using large-scale reinforcement learning and multi-agent reasoning, Asimov can plan, write, and review code autonomously. Within its first year, the company attracted top-tier investors including Sequoia, CRV, and Lightspeed, raising more than $130 million and scaling to a team of over 30 engineers across New York, San Francisco, and London.

Throughout his career, Ioannis has remained driven by a simple but powerful motivation: to work on ideas that matter. His approach blends deep research expertise with a pragmatic focus on reliability, safety, and real-world impact. He advocates for AI systems that are aligned with human values, open rather than closed, and capable of amplifying human capability rather than replacing it.

From a young researcher in Greece to a founding engineer behind AI systems that changed the world - and now a founder building the next generation of intelligent agents - Ioannis Antonoglou embodies the mindset that defines high-impact entrepreneurship: bold ideas, long-term vision, and relentless dedication to solving hard problems.