# Christian Szegedy

> Hungarian mathematician and AI researcher Christian Szegedy co-authored landmark work on adversarial examples, Inception and batch normalization at Google, joined xAI’s founding team, and now leads AletheAI as co-founder and CEO while pursuing machine-verifiable mathematical reasoning.

- **Role:** Co-founder and CEO at AletheAI
- **Organizations:** AletheAI, Math Inc., Morph Labs, xAI, Google Research, Cadence Design Systems
- **Nationality:** Hungarian
- **Education:** PhD in Mathematics, University of Bonn, Mathematics studies, Eötvös Loránd University
- **Known for:** First author of the GoogLeNet/Inception paper behind the 2014 ImageNet classification win., Co-author of the 2015 batch normalization paper, which received the ICML 2025 Test of Time Award., First author of Intriguing Properties of Neural Networks, a foundational study of adversarial examples.

## Career timeline

- **2005** — Completed a mathematics doctorate at the University of Bonn on the weighted combinatorial Laplacian.
- **2005–2010** — Worked on chip-design research at Cadence Design Systems.
- **2010–2023** — Worked at Google, first in ad optimization and later in computer vision and mathematical reasoning research.
- **2013** — First-authored Intriguing Properties of Neural Networks on adversarial examples.
- **2014–2015** — Led the GoogLeNet/Inception paper associated with the 2014 ImageNet classification win.
- **2015** — Co-authored Batch Normalization with Sergey Ioffe.
- **2016** — Co-authored DeepMath and began the N2Formal research program.
- **2023–2025** — Joined xAI’s founding team.
- **2025** — Worked as chief scientist at Morph Labs and founded Math Inc.
- **2026** — Listed by AletheAI as co-founder and CEO; spoke at Bonn on the automatic mathematician.

## Achievements

- First author of the GoogLeNet/Inception paper behind the 2014 ImageNet classification win.
- Co-author of the 2015 batch normalization paper, which received the ICML 2025 Test of Time Award.
- First author of Intriguing Properties of Neural Networks, a foundational study of adversarial examples.
- Co-author of DeepMath, an early large-scale application of deep learning to theorem proving.
- Founded Google’s N2Formal team in 2016.

## Latest updates

- **2026-05** — Spoke at the University of Bonn on the history and near future of automatic mathematics; the event identified him with AletheAI.
- **2026-05** — Discussed xAI, Math Inc. and machine-checkable reasoning on The Information Bottleneck podcast.
- **2025-09** — Announced Math Inc. and its Gauss autoformalization work.

## Links

- Website: https://alethe-ai.com/
- LinkedIn: https://www.linkedin.com/in/christian-szegedy-bb284816
- Twitter/X: https://x.com/ChrSzegedy

---

Profile page: https://yespress.io/christian-szegedy
Published by YesPress — https://yespress.io
Last updated: 2026-09-29