Karén Simonyan

Karén Simonyan trained in computer vision at the University of Oxford's Visual Geometry Group under Andrew Zisserman. In 2014 the two published "Very Deep Convolutional Networks for Large-Scale Image Recognition," which showed that stacking many small 3-by-3 convolution filters, to depths of 16 and 19 layers, improved image recognition. The networks, known as VGG after the group, placed first in localization and second in classification at the 2014 ImageNet challenge, and their released weights became a standard starting point for other vision systems. The same year Simonyan and Zisserman proposed two-stream networks, which recognize actions in video by processing still frames and motion separately.

With Zisserman and Max Jaderberg, Simonyan co-founded Vision Factory, an Oxford spin-out aiming at "unconstrained visual recognition of objects, actions, and text." Google DeepMind hired the three in October 2014. At DeepMind Simonyan was a co-author of WaveNet (2016), which generated speech one audio sample at a time, of AlphaGo Zero and AlphaZero, which learned Go, chess and shogi from self-play alone, and of BigGAN (2018), a large image generator.

In March 2022 Simonyan co-founded Inflection AI with Mustafa Suleyman and Reid Hoffman and became chief scientist of the company, which built the Pi chatbot. When Microsoft hired most of Inflection's staff in March 2024, Satya Nadella named Simonyan chief scientist of the new Microsoft AI, reporting to Suleyman, and credited Simonyan with leading "some of the biggest AI breakthroughs over the past decade including AlphaZero." In November 2025 Simonyan became chief scientist of the MAI Superintelligence Team. The organization released its first in-house models in August 2025 and seven more in June 2026, including the reasoning model MAI-Thinking-1, which Microsoft AI said it trained from scratch without distilling other labs' models.