Türkiye-born computer scientist Koray Kavukcuoglu has been handed one of the most influential jobs in global technology, taking over the day-to-day leadership of Google DeepMind as the tech giant intensifies its race to develop artificial general intelligence (AGI).
Google and Alphabet chief executive Sundar Pichai announced last week that Kavukcuoglu has been promoted to Senior Vice President (SVP) of Google DeepMind, while retaining his position as Chief AI Architect at Google and reporting directly to Pichai.
The appointment puts the Turkish engineer in charge of Gemini model development, Google DeepMind’s Frontier AI research, as well as its Gemini app and developer teams – giving him enormous influence over both the development of Google’s most advanced AI technology and how that technology reaches billions of users.
Google itself describes Kavukcuoglu as “one of the world's foremost experts in artificial intelligence”, while Pichai highlighted his 13-year history at DeepMind and his role in establishing its deep learning team and leading breakthroughs including Deep Q-Network (DQN) and WaveNet.
The promotion also marks an extraordinary rise for an engineer whose career began in Türkiye and whose first university discipline was not computer science, but aerospace engineering.
A Turkish engineer at the summit of global AI
Born in Türkiye, Kavukcuoglu studied Aerospace Engineering at Ankara’s prestigious Middle East Technical University (‘ODTÜ’), completing his undergraduate degree in 1999 and a master’s in the same field in 2003.
During this period, he also spent around four years working as a research and development engineer for Turkish defence technology company Roketsan, giving him an important early professional connection to Türkiye’s high-tech engineering sector.
He subsequently moved to the United States, where he switched his focus to computer science at New York University (NYU). There he completed postgraduate studies, including his PhD, working within the Computational and Biological Learning Lab under Yann LeCun, one of the pioneering figures in modern deep learning.
After working at NEC Laboratories America, Kavukcuoglu joined DeepMind in 2012, while it was still an independent British AI start-up and two years before it was acquired by Google.
He went on to found DeepMind’s deep learning team and became a central figure in research that helped establish the organisation at the forefront of modern artificial intelligence.
Demis Hassabis on Koray Kavukcuoglu: “one of the world's foremost AI experts”
Among the breakthroughs associated with teams he led is DQN, or Deep Q-Network, which combined deep neural networks with reinforcement learning to enable an AI system to learn how to play Atari video games directly from screen images and rewards. Kavukcuoglu was one of the lead authors of the landmark research.
His teams also pioneered WaveNet, a deep neural network capable of generating highly realistic speech, and he has contributed to wider DeepMind research, including work connected to AlphaFold. Google later put Kavukcuoglu at the heart of the development of its Gemini family of generative AI models.
His standing in the international engineering community was further recognised in 2022 when he was elected a Fellow of Britain’s Royal Academy of Engineering for his contributions and leadership in areas including deep learning, generative models, reinforcement learning and the application of AI to real-world problems.
Taking over from Demis Hassabis
Kavukcuoglu’s new position is especially significant because it sees him assume the operational leadership previously held by Google DeepMind co-founder Sir Demis Hassabis, one of the best-known figures in artificial intelligence.
Hassabis is not leaving Google DeepMind. Instead, the British AI pioneer and Nobel laureate is becoming its Chair and Chief Scientist of Alphabet, allowing him to step away from day-to-day management and concentrate on longer-term questions surrounding AGI, science and the impact of increasingly powerful artificial intelligence.
“I’ve decided that now is the right time for me to hand over my day-to-day operational responsibilities at GDM,” Hassabis told employees, saying he wanted the “time and space to focus on the big picture”.
An emphatic endorsement
Hassabis gave his successor an emphatic endorsement, stating that he and Kavukcuoglu had worked together for more than 13 years. He described him as “one of the world's foremost AI experts”, and that he had “total confidence” in Kavukcuoglu and the wider Google DeepMind executive team.
The change therefore represents more than a reshuffling of titles. Kavukcuoglu now becomes the executive with ultimate day-to-day responsibility for turning Google DeepMind’s research into the next generation of AI models and products.
Reuters reported this week that Google had confirmed Kavukcuoglu will have the final say on major DeepMind decisions. His influence had already been growing significantly: when Pichai created the role of Chief AI Architect for him in 2025, Kavukcuoglu relocated from London to Google’s headquarters in Mountain View, California, with a mandate to accelerate the use of DeepMind models throughout Google's products.
What comes next for Gemini and Google AI?
The most immediate priority will be Gemini, Google's answer to rival AI systems from the likes of OpenAI and Anthropic.
Kavukcuoglu was already closely involved in developing Gemini 3 and its successors. Google released Gemini 3.5 earlier this year with a particular emphasis on “agentic” AI – systems capable of carrying out more complex sequences of tasks rather than merely responding to individual prompts.
Hassabis has now confirmed that work is under way on Gemini 4, saying Google's flagship models are “in good hands with Koray and the leads” and pointing to progress being made on the next generation. Google has not announced a release date.
Kavukcuoglu’s combined responsibilities also give an indication of where Google's AI is heading.
As Chief AI Architect, he was specifically charged with bringing the company's most advanced models into consumer products more rapidly and efficiently. Pichai said when appointing him in 2025 that the goal was more seamless integration, quicker iteration and greater efficiency across Google.
That suggests the boundary between Google DeepMind as a research laboratory and Google as a consumer technology company will become increasingly thin.
Kavukcuoglu has previously argued that having AI research and products closely connected creates a powerful feedback loop: researchers can turn breakthroughs into tools used at enormous scale, while the ways consumers and developers use those products can in turn highlight where the underlying models need to improve.
Speaking to the Financial Times earlier this year, he placed particular emphasis on AI becoming more agentic, able to reason about a problem and decide which tools or actions are required to solve it, while learning from the increasingly demanding ways people use the technology.
Google DeepMind's recent work points in the same direction. Alongside increasingly capable Gemini models, it is developing multimodal systems able to work across text, images, sound and video; AI agents for scientific research; and Gemini-powered robotics capable of understanding their physical surroundings and planning multi-step tasks.
All of this remains tied to DeepMind’s ultimate and far more ambitious mission: the development of artificial general intelligence, or AGI – AI capable of performing a very broad range of intellectual tasks at, or potentially beyond, human levels.
High expectations – and intense competition
Kavukcuoglu also inherits the job at a testing moment.
Google is competing intensely with OpenAI, Anthropic and others to remain at the frontier of generative AI, while simultaneously trying to turn the huge sums being invested in artificial intelligence infrastructure and research into successful mass-market products.
Reuters reported that Google has been pushing DeepMind to increase the pace of development and that the company has faced challenges around computing capacity, organisational complexity and the speed with which new Gemini models reach users. The reorganisation has also prompted questions internally about how much independence DeepMind will retain as it becomes more tightly woven into Google.
That may be precisely why Kavukcuoglu has emerged as such a pivotal figure.
Unlike an executive parachuted in to commercialise DeepMind's research, he has been part of the organisation since its earliest years and helped create some of the research on which its reputation was built. At the same time, his role as Google's Chief AI Architect has made him a key bridge between frontier research and products used by consumers and businesses.
For Türkiye, his appointment is also a remarkable milestone: an ODTÜ-trained aerospace engineer who began his professional life developing technology at Roketsan is now directing the AI research and model development of one of the world's most powerful technology companies.
And for Kavukcuoglu himself, the promotion cements a journey from Turkish engineering graduate and young AI researcher to one of the handful of people who will have a major say in determining how artificial intelligence develops – and how billions of people ultimately experience it.
