PeopleofSuperintelligence

Researcher

Andrew Howard

Also known as Andrew G. Howard

Andrew Howard is a computer vision researcher whose work develops compact image-recognition models for mobile devices. His early research includes a 2004 kernel paper and a 2007 method for learning classification transforms. He helps release MobileNets in 2017, coauthors MobileNetV2 in 2018, and coauthors MobileNetV3 in 2019. By 2023, he had completed a computer-science PhD at Columbia and led applied researchers as a Senior Staff Software Engineer at Google Research. He coauthors MobileNetV4 in 2024, which introduces blocks designed for efficient mobile hardware.

Now at
Google
Works
33
Citations
68,042
h-index
19
Rankings
#75 by citations since 2021
Badges

Publication numbers from OpenAlex, as of 18 October 2026.

Timeline

  1. 2004–2007

    Probability-product kernels

    In 2004, Howard coauthors Probability Product Kernels with Risi Kondor and Tony Jebara, combining discriminative learning with generative modeling. In 2007, he and Jebara present a method that learns monotonic transformations during classification.

  2. June 2017

    MobileNets for on-device vision

    In June 2017, Howard and his coauthors release MobileNets as small, low-latency TensorFlow models for image classification, detection, embeddings and segmentation on mobile and embedded devices. Depthwise separable convolutions and width and resolution multipliers let developers tune models for device constraints.

  3. 2018

    The MobileNetV2 upgrade

    In 2018, Howard and Mark Sandler publish MobileNetV2 for mobile image classification, object detection and semantic segmentation. Its inverted residual blocks use linear bottlenecks and shortcut connections, reducing operations and parameters compared with MobileNetV1.

  4. November 2019

    MobileNetV3 and Edge TPU

    Howard and Suyog Gupta introduce MobileNetV3 and MobileNetEdgeTPU in November 2019 for on-device vision. MobileNetV3 combines neural architecture search with NetAdapt for mobile CPUs, while MobileNetEdgeTPU tunes models for Pixel 4's Edge TPU.

  5. 2023–2024

    Senior staff engineer at Google Research

    By 2023, Howard has a computer-science PhD from Columbia University focused on machine learning and is a Senior Staff Software Engineer at Google Research. He leads a team of applied researchers working on efficient computer vision for mobile applications. In 2024, he coauthors MobileNetV4, which introduces a Universal Inverted Bottleneck block and Mobile MQA for mobile accelerators.

Career

  1. 2016–2024Google
  2. Google ResearchSenior Staff Software Engineer

Education

  1. Columbia UniversityPhD

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 · 28,580 citations
  2. MobileNets: Efficient Convolutional Neural Networks for Mobile Vision Applications2017 · 19,001 citationsWith Hartwig Adam
  3. Searching for MobileNetV32019 · IEEE/CVF International Conference on Computer Vision (ICCV) · 9,596 citationsWith Hartwig Adam, Quoc Le and Vijay Vasudevan
  4. Quantization and Training of Neural Networks for Efficient Integer-Arithmetic-Only Inference2018 · IEEE Computer Society Conference on Computer Vision and Pattern Recognition · 4,066 citationsWith Hartwig Adam
  5. MnasNet: Platform-Aware Neural Architecture Search for Mobile2019 · IEEE Computer Society Conference on Computer Vision and Pattern Recognition · 3,406 citationsWith Quoc Le and Vijay Vasudevan

From OpenAlex, as of 18 October 2026.

  1. Google has the highest combined h-index among AI organizationsIncludes Google. Our estimates of combined and median h-index at AI organizations, using current affiliations and OpenAlex records.
  2. Google alumni founded or led 82 other organizations, the most among the labs we compareIncludes Google. We trace 189 organizations founded or led by AI lab alumni, and 196 career moves between the labs.
  3. 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 from 1961 to 2025, with a year-by-year chart.
  4. Of 425 people shaping AI who worked at Google, 173 work there nowIncludes Google. Our estimates of where former Google and Google DeepMind people work now, including moves within Google.

Colleagues

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

Sources and credits

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

  1. Probability Product Kernels jmlr.org. Checked 18 October 2026.
  2. Learning Monotonic Transformations for Classification papers.nips.cc. Checked 18 October 2026.
  3. ECV23 - Invited talks sites.google.com. Checked 18 October 2026.
  4. MobileNets: Open Source Models for Efficient On-Device Vision opensource.googleblog.com. Checked 18 October 2026.
  5. MobileNets: Efficient Convolutional Neural Networks for Mobile Vision Applications arxiv.org. Checked 18 October 2026.
  6. MobileNetV2: The Next Generation of On-Device Computer Vision Networks research.google. Checked 18 October 2026.
  7. MobileNetV2: Inverted Residuals and Linear Bottlenecks arxiv.org. Checked 18 October 2026.
  8. Introducing the Next Generation of On-Device Vision Models: MobileNetV3 and MobileNetEdgeTPU research.google. Checked 18 October 2026.
  9. Searching for MobileNetV3 arxiv.org. Checked 18 October 2026.
  10. MobileNetV4 -- Universal Models for the Mobile Ecosystem arxiv.org. Checked 18 October 2026.