Christian Szegedy
Co-Founder & CEO
Christian Szegedy is one of the most influential researchers in modern deep learning. At Google Brain, he created the GoogLeNet/Inception architecture that won the 2014 ImageNet challenge, co-invented batch normalization — one of the most widely adopted techniques in all of deep learning — and authored the foundational paper on adversarial examples. His work has been cited over 200,000 times.
Since 2016, Christian has pioneered the application of AI to mathematics and formal reasoning. His DeepMath project was the first to demonstrate that neural networks could be effective for large-scale theorem proving. He has since developed a vision for autoformalization — AI systems that translate informal mathematics into machine-verifiable formal proofs — as a path toward verified, trustworthy AI. He holds a PhD in Mathematics from the University of Bonn.
Michael R. Douglas
Co-Founder & Chief Scientist
Michael R. Douglas is a theoretical physicist who has worked across string theory, mathematical physics, computational physics, and machine learning. He received his BA from Harvard and his PhD from Caltech under John Schwarz, one of the creators of string theory. At Caltech he also studied computation and physics with John Hopfield, Richard Feynman, and Gerald Sussman.
As a professor at Rutgers University, where he directed the New High Energy Theory Center from 2000 to 2007, he made major contributions to the second superstring revolution. In 2008 he became the first permanent member of the Simons Center for Geometry and Physics at Stony Brook. From 2012 to 2020 he was a researcher at Renaissance Technologies. He is currently a senior research scientist at the CMSA, Harvard University, developing new ways to use machine learning in mathematical research, and a PI in the Simons Collaboration on Physics of Learning and Neural Computation.