Koray Kavukcuoglu

Koray Kavukcuoglu trained as an aerospace engineer before taking a PhD in computer science at New York University in Yann LeCun's group, where he worked on unsupervised learning and multi-stage architectures for object recognition. After a spell as a researcher at NEC Labs America he joined DeepMind in 2012, two years before Google bought it. He founded the company's deep learning team, which produced the Deep Q-Network that learned Atari games from pixels, the IMPALA reinforcement learning architecture and WaveNet, the speech-generation model that later voiced Google Assistant.

As vice president of research he led the Scientific Board that Hassabis set up in April 2023 to oversee research direction when DeepMind merged with Google Brain. By 2025 he was Google DeepMind's chief technology officer, and in June 2025 Sundar Pichai made him Google's first Chief AI Architect, a senior vice president post reporting to Pichai, with a brief to "accelerate how we bring our world-leading models into our products." He moved from London to Mountain View for the role. He and Hassabis were the program leads named in the December 2023 Gemini technical report, and he co-signed the November 2025 launch post for Gemini 3 with Pichai and Hassabis.

On 5 August 2026 Pichai announced that Kavukcuoglu would run Google DeepMind as senior vice president, overseeing Gemini model development, frontier research and the Gemini app and developer teams, while Hassabis became Chair. He took over a lab whose Gemini 3.5 Pro, announced in May, had missed its June release, and whose chief scientist Jeff Dean left the same day. At The Information's AI Agenda Live summit in September 2026, his first public appearance in the job, he said Gemini 4 was in post-training and that Google intended to release an early version "as soon as possible."