Christian Szegedy
Christian Szegedy is an AI researcher whose work spans computer vision, neural-network training and formal mathematics. He completed a mathematics doctorate at Bonn in 2005, worked at Cadence Design Systems, and joined Google as a software engineer in 2010. At Google, he coauthored papers on adversarial examples, Inception and batch normalization, then helped build neural theorem proving and autoformalization research. He joined xAI's founding team, became chief scientist at Morph Labs in 2025, and was at Math Inc by May 2026; ICLR and ICML honored two papers he coauthored in 2024 and 2025. In May 2026, he was affiliated with AletheAI for a Bonn talk, and in September he appeared among AITP 2026 participants and speakers.
- Now at
- xAI
- Works
- 68
- Citations
- 168,012
- h-index
- 26
Publication numbers from OpenAlex, as of 14 October 2026.
Timeline
-
2005–2010
From Bonn to Cadence
He completes a mathematics doctorate at the University of Bonn in 2005. He works as a research scientist at Cadence Design Systems from 2005 to 2010 and joins Google as a software engineer in 2010.
-
2013–2015
Inception and batch normalization
In 2013, he coauthors Intriguing Properties of Neural Networks, showing that imperceptible image changes can fool a classifier. His 2015 CVPR paper Going Deeper with Convolutions introduces Inception for ImageNet classification and detection. With Sergey Ioffe, he publishes Batch Normalization in 2015, which supports higher learning rates and less careful initialization.
-
2016–2020
The N2Formal program
He starts Google's N2Formal team in 2016 and coauthors DeepMath, which uses neural sequence models to select premises for automated theorem proving. In 2020, he argues that autoformalization can help systems learn general reasoning in mathematics and computer science.
-
2023
Joining xAI
In 2023, he joins xAI's founding team after 12 years as a Google engineer and research scientist.
-
2024–2025
Test-of-time awards
ICLR names his 2013 paper Intriguing Properties of Neural Networks a Test of Time runner-up in 2024. In 2025, ICML gives its Test of Time Award to Batch Normalization, coauthored with Sergey Ioffe.
-
2025
Chief scientist at Morph
In 2025, he joins Morph Labs as chief scientist after his xAI co-founder role.
-
May 2026
Gauss at Math Inc
By May 2026, he is at Math Inc and works on formal mathematics. Its Gauss system completes a proof in two weeks after human experts estimated that the work would take another year.
-
May 2026–September 2026
An AITP speaker
In May 2026, he is affiliated with AletheAI for a Bonn talk on autoformalization and machine-verifiable mathematics. In September 2026, he appears among AITP 2026 participants and speakers.
Sources and credits
Photo: We have no published photo of Christian Szegedy yet. Send us a free-licensed photo.
- Christian Szegedy - The Mathematics Genealogy Project
- Christian Szegedy | DeepAI
- Intriguing properties of neural networks
- Going Deeper With Convolutions
- Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift
- Towards the Automatic Mathematician: The Past, The Present, and the Next Few Months Ahead (Yes, Really)
- DeepMath - Deep Sequence Models for Premise Selection
- A Promising Path Towards Autoformalization and General Artificial Intelligence
- Elon Musk Announces xAI and Its 12-Man Team: Here’s Who They Are
- ICLR 2024 Test of Time Award
- ICML 2025 Awards
- The infrastructure for the singularity
- Inside xAI, and the Bet on AI Math - with Christian Szegedy (Math Inc)
- AITP'26: 11th Conference on Artificial Intelligence and Theorem Proving