ABAB Pedia

Alexey Dosovitskiy

Alexey Dosovitskiy is a computer vision and machine learning researcher and a RecursiveRecursiveRecursive (Recursive Superintelligence) is a company pursuing automated AI research and recursive self-improvement, co-founded by Richard Socher and seven others. It emerged publicly in May 2026, released training and GPU-kernel optimization experiments, and signed a multi-year compute agreement with AWS.Open the full entry co-founder, with contributions to FlowNet, CARLA and Vision Transformer.

Contents7 sections
Key facts

Mathematical training

Dosovitskiy earned a master’s-level mathematics degree and a PhD at Moscow State University in 2009 and 2012, before moving into computer vision.⁠[1]

Research in Freiburg

He conducted postdoctoral research in Thomas Brox’s computer vision group at the University of Freiburg, studying unsupervised learning, image generation and motion estimation.⁠[1]

FlowNet

The FlowNet team treated optical-flow estimation as supervised learning, using convolutional networks to estimate motion between images and creating synthetic Flying Chairs training data. Dosovitskiy co-authored the paper.⁠[2]

CARLA driving simulator

He led the CARLA paper introducing an open-source urban driving simulator with configurable sensors and environmental conditions for developing, training and validating autonomous-driving systems.⁠[3]

Google and Vision Transformer

Biology and Recursive

Sources

  1. Alexey Dosovitskiy (opens in a new window)
  2. FlowNet: Learning Optical Flow with Convolutional Networks (opens in a new window)
  3. CARLA: An Open Urban Driving Simulator (opens in a new window)
  4. An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale (opens in a new window)
  5. Recursive Superintelligence: Why Self-Improving AI is the Next Frontier (opens in a new window)