PeopleofSuperintelligence

Researcher

Mark Sandler

Mark Sandler is a Google researcher whose AI work spans collaborative filtering, mobile computer vision and few-shot learning. He coauthored a collaborative-filtering paper in 2003 and work on unsupervised text classification in 2005, then studied sparse convolutional networks in 2017. MobileNetV2, MnasNet and MobileNetV3 brought his work to efficient vision on phones; later papers explored model compression, mobile object detection and few-shot learning. His work also included COVID-19 mobility modeling and task-specific transformers, along with a 2026 paper on robust training across precision and sparsity. As of October 2026, he is a Google researcher in machine intelligence and machine perception.

Works
15
Citations
43,021
h-index
9
Badges

Publication numbers from OpenAlex, as of 22 October 2026.

Timeline

  1. 2003–2017

    Collaborative filtering and sparse CNNs

    In 2003, he coauthors “Convergent algorithms for collaborative filtering” with Jon M. Kleinberg at the ACM Conference on Electronic Commerce. At KDD in 2005, he publishes “On the use of linear programming for unsupervised text classification.” In 2017, he coauthors “The Power of Sparsity in Convolutional Neural Networks,” which tests channel reduction to balance model size, speed and accuracy.

  2. 2018

    MobileNetV2 for mobile vision

    In 2018, he coauthors MobileNetV2, which introduces linear bottlenecks and shortcut connections between bottlenecks. With SSDLite, the model is about 35% faster than MobileNetV1 at the same object-detection accuracy.

  3. 2019

    MnasNet's phone-latency search

    In 2019, he coauthors MnasNet, a mobile neural-architecture search method that measures inference latency directly on phones. On ImageNet, it reaches 75.2% top-1 accuracy at 78 ms on a Pixel phone.

  4. 2019

    MobileNetV3 and model compression

    In 2019, he coauthors Searching for MobileNetV3, which combines hardware-aware architecture search with NetAdapt to tune models for mobile CPUs. He also coauthors Structured Multi-Hashing for Model Compression, which combines weight hashing and dimensionality reduction to control deep-network size.

  5. 2020–2022

    SpotPatch and HyperTransformer

    In 2020, he coauthors SpotPatch, which adapts mobile object detectors by sharing model weights and adding a small task-specific patch. He also coauthors a Bayesian analysis that uses mobile-phone mobility data to estimate COVID-19 infection-rate changes and forecast deaths. In 2022, he coauthors HyperTransformer for generating CNN weights from few-shot examples and a CVPR study of learnable memory tokens for image transformers.

  6. 2024

    Continual HyperTransformer

    In 2024, he coauthors Continual HyperTransformer, which reuses generated CNN weights across tasks so the model can learn sequential few-shot tasks without forgetting earlier tasks.

  7. 2025

    Projectable Models

    In 2025, he coauthors Projectable Models, which maps parameters of a large Transformer to a smaller model tailored to a specific task.

  8. As of October 2026

    Google Research at ICLR 2026

    In 2026, he coauthors Robust Training of Neural Networks at Arbitrary Precision and Sparsity at ICLR. As of October 2026, he is a Google researcher in machine intelligence and machine perception.

Career

  1. NowGoogle ResearchResearcher
  2. 2017–2021Google

From the biography and OpenAlex affiliations (the years of papers that list the organization), so some years are approximate.

Most-cited works

  1. MobileNetV2: Inverted Residuals and Linear Bottlenecks2018 · IEEE Computer Society Conference on Computer Vision and Pattern Recognition · 28,599 citationsWith Andrew Howard and Menglong Zhu
  2. Searching for MobileNetV32019 · IEEE/CVF International Conference on Computer Vision (ICCV) · 9,605 citationsWith Andrew Howard, Hartwig Adam, Quoc Le and Vijay Vasudevan
  3. MnasNet: Platform-Aware Neural Architecture Search for Mobile2019 · IEEE Computer Society Conference on Computer Vision and Pattern Recognition · 3,407 citationsWith Andrew Howard, Quoc Le and Vijay Vasudevan
  4. Inverted Residuals and Linear Bottlenecks: Mobile Networks for Classification, Detection and Segmentation2018 · 671 citationsWith Andrew Howard and Menglong Zhu
  5. NetAdapt: Platform-Aware Neural Network Adaptation for Mobile Applications2018 · Lecture notes in computer science · 572 citationsWith Andrew Howard, Hartwig Adam and Vivienne Sze

From OpenAlex, as of 21 October 2026.

  1. Google had the most people shaping AI every year from 2014 to 2025Includes Google. Our estimates of which organizations employed the most people shaping AI each year from 1955 to 2025, tracing the shift from university labs to companies.
  2. Google has the highest combined h-index among AI organizationsIncludes Google. We estimate that Google holds the highest combined h-index at 8,016, while universities account for 73% of the deduplicated total across AI organizations.
  3. 20% of people at companies who shape AI also hold a university role, led by Google and Stanford UniversityIncludes Google. We estimate that 539 of the 6,766 people with a known current organization (8%) hold both a university role and a company role, including 20% of people at companies and 13% of people at universities.
  4. Google alumni have founded or led 133 other organizations, more than any other AI labIncludes Google. We estimate that alumni of 17 major AI labs went on to found or lead 350 other organizations, including 226 they founded, and we trace 472 career moves directly between 23 labs since 2010.

Colleagues

People with dated positions at the same organization in the same years.

Sources and credits

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

  1. Convergent algorithms for collaborative filtering research.google. Checked 20 October 2026.
  2. On the use of linear programming for unsupervised text classification research.google. Checked 20 October 2026.
  3. KDD 2005: Chicago, Illinois, USA sigmod.org. Checked 20 October 2026.
  4. The Power of Sparsity in Convolutional Neural Networks arxiv.org. Checked 20 October 2026.
  5. MobileNetV2: The Next Generation of On-Device Computer Vision Networks research.google. Checked 20 October 2026.
  6. MnasNet: Platform-Aware Neural Architecture Search for Mobile research.google. Checked 20 October 2026.
  7. Searching for MobileNetV3 arxiv.org. Checked 20 October 2026.
  8. Structured Multi-Hashing for Model Compression arxiv.org. Checked 20 October 2026.
  9. Mark Sandler research.google. Checked 20 October 2026.
  10. Continual HyperTransformer: A Meta-Learner for Continual Few-Shot Learning research.google. Checked 20 October 2026.
  11. Projectable Models: One-Shot Generation of Small Specialized Transformers from Large Ones arxiv.org. Checked 20 October 2026.
  12. Google at ICLR 2026 research.google. Checked 20 October 2026.