PeopleofSuperintelligence

Researcher

Sergey Ioffe

Also known as S. Ioffe

Sergey Ioffe is a computer-vision researcher whose work helped speed deep-network training. At Berkeley, he completed a computer-science PhD with a thesis on object recognition, then held research roles at Fujifilm and Google. His Google work includes MultiBox object detection and efficient Inception networks. In 2015, he and Christian Szegedy published Batch Normalization, which lets deep networks use higher learning rates and train in fewer steps; ICML gave the paper its 2025 Test of Time Award. He introduced Batch Renormalization in 2017 and is head of AI at Scaled Foundations, a role he began in 2025.

Now at
Scaled Foundations
Works
33
Citations
87,885
h-index
23

Publication numbers from OpenAlex, as of 15 October 2026.

Timeline

  1. 1996–2001

    A Berkeley PhD in computer science

    He studies computer science at the University of California, Berkeley, from 1996 to 2001 and completes a PhD thesis on managing correspondence search in object recognition in 2001.

  2. 2001–2006

    Research scientist at Fujifilm

    He works as a research scientist at Fujifilm from 2001 to 2006.

  3. 2006–2024

    Senior staff scientist at Google

    He works as a senior staff research scientist at Google from 2006 to 2024.

  4. 2014–2015

    Detection and Inception networks

    He coauthors the 2014 Scalable, High-Quality Object Detection paper, which presents MultiBox as a learning-based object-proposal method with efficient runtime-quality trade-offs. In 2015, he coauthors Rethinking the Inception Architecture for Computer Vision, which uses factorized convolutions and aggressive regularization to make image-recognition networks more efficient.

  5. 2015

    Batch Normalization

    He and Christian Szegedy publish Batch Normalization in 2015. The method lets deep networks use higher learning rates and reach the same image-classification accuracy with 14 times fewer training steps.

  6. 2017

    Batch Renormalization

    He introduces Batch Renormalization in 2017 to improve model training when minibatches are small or contain non-independent samples.

  7. 2025

    An ICML Test of Time Award

    ICML gives him and Christian Szegedy its 2025 Test of Time Award for Batch Normalization.

  8. Since 2025

    Head of AI at Scaled Foundations

    He takes the role of head of AI at Scaled Foundations in 2025.

Sources and credits

Photo: We have no published photo of Sergey Ioffe yet. Send us a free-licensed photo.

  1. UC Berkeley Computer Vision Group www2.eecs.berkeley.edu. Checked 15 October 2026.
  2. Sergey Ioffe | alphaXiv alphaxiv.org. Checked 15 October 2026.
  3. Scalable, High-Quality Object Detection arxiv.org. Checked 15 October 2026.
  4. Rethinking the Inception Architecture for Computer Vision arxiv.org. Checked 15 October 2026.
  5. Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift proceedings.mlr.press. Checked 15 October 2026.
  6. Batch Renormalization: Towards Reducing Minibatch Dependence in Batch-Normalized Models arxiv.org. Checked 15 October 2026.
  7. ICML 2025 Awards icml.cc. Checked 15 October 2026.