PeopleofSuperintelligence

Researcher

Mingxing Tan

Mingxing Tan is a machine-learning researcher whose work spans efficient vision models and autonomous-driving systems. He earned a computer-science PhD with highest honors at Peking University in 2013 and worked as a postdoctoral associate at Cornell through 2015. His later work covers EfficientNet, EfficientDet, EfficientNetV2, 3D point-cloud detection, multimodal driving models and a benchmark for rare road scenarios. As of October 2026, he is a Research Director at Waymo AI and leads its foundation and world-model teams.

Now at
Waymo
Badges

Timeline

  1. 2013–2015

    A Peking PhD and Cornell postdoc

    He earns a computer science PhD with highest honors at Peking University in 2013 after working as a research assistant and leading the Microprocessor Research and Development Center's compiler group. From 2013 to 2015, he is a postdoctoral associate in Cornell's School of Electrical and Computer Engineering, working in the Computer Systems Lab with Zhiru Zhang.

  2. 2019

    EfficientNet compound scaling

    In 2019, he and Quoc Le propose compound scaling, which balances network depth, width and resolution through one coefficient, and use neural architecture search to build the EfficientNet family. EfficientNet-B7 reaches 84.4% top-1 accuracy on ImageNet while being 8.4 times smaller and 6.1 times faster at inference than the comparison ConvNet.

  3. 2020

    EfficientDet object detection

    In 2020, he and Adams Yu introduce EfficientDet, a scalable object detector built with an EfficientNet backbone and a bidirectional feature network. EfficientDet-D7 reaches 52.2 mAP on COCO with four times fewer parameters and 9.4 times less computation than the previous state of the art.

  4. 2021

    EfficientNetV2 trains faster

    In 2021, he and Zihang Dai introduce EfficientNetV2, a convolutional model family built with training-aware neural architecture search and progressive learning. On ImageNet, it trains about 5–11 times faster, with a model size up to 6.8 times smaller than previous models and no accuracy drop.

  5. 2022

    SWFormer detects point clouds

    In 2022, he coauthors SWFormer, a sparse-window transformer for 3D object detection that turns point clouds into sparse voxels and windows. On the Waymo Open Dataset's official test set, it reaches 73.36 L2 mAPH for vehicles and pedestrians.

  6. 2024

    EMMA for autonomous driving

    In 2024, he coauthors EMMA, an end-to-end multimodal model built on a large language model foundation such as Gemini. It maps camera data into planning trajectories, perception objects and road-graph elements, and achieves state-of-the-art motion planning on nuScenes with competitive results on the Waymo Open Motion and Waymo Open datasets.

  7. 2025

    A dataset for rare scenarios

    In 2025, he coauthors WOD-E2E, the Waymo Open Dataset for End-to-End Driving, a benchmark with 4,021 segments curated for long-tail scenarios that occur less than 0.03% of the time in daily driving. It introduces the Rater Feedback Score, which compares predicted trajectories with rater-annotated preferences.

  8. As of October 2026

    Research director at Waymo

    As of October 2026, he is a Research Director at Waymo AI and leads its foundation and world-model teams. He helps define the Waymo World Model's vision and coauthors Sensor2Sensor, a 2026 CVPR paper on cross-embodiment sensor conversion for autonomous driving.

Career

  1. NowWaymoResearch Director
  2. 2013–2015Cornell UniversityPostdoctoral Associate
  3. Microprocessor Research and Development CenterResearch Assistant and compiler group lead

Education

  1. 2013Peking UniversityPhD

From the biography.

  1. 16% of AI career moves in the 2020s went from a university to a companyIncludes Mingxing Tan. We estimate that 16% of AI career moves in the 2020s went from a university to a company; 29% of moves starting at a university went to a company, and 76% ended the university role.
  2. 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.
  3. 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.
  4. 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.

Colleagues

People named in this biography, and people with dated positions at the same organization in the same years.

Sources and credits

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

  1. ASCENT Fellows katesnyderdavis.wixsite.com. Checked 21 October 2026.
  2. People zhang.ece.cornell.edu. Checked 21 October 2026.
  3. EfficientNet: Rethinking Model Scaling for Convolutional Neural Networks proceedings.mlr.press. Checked 21 October 2026.
  4. EfficientDet: Towards Scalable and Efficient Object Detection research.google. Checked 21 October 2026.
  5. Toward Fast and Accurate Neural Networks for Image Recognition research.google. Checked 21 October 2026.
  6. SWFormer: Sparse Window Transformer for 3D Object Detection in Point Clouds arxiv.org. Checked 21 October 2026.
  7. EMMA: End-to-End Multimodal Model for Autonomous Driving arxiv.org. Checked 21 October 2026.
  8. WOD-E2E: Waymo Open Dataset for End-to-End Driving in Challenging Long-tail Scenarios arxiv.org. Checked 21 October 2026.
  9. DriveX 2026 – Foundation Models for Autonomous Driving drivex-workshop.github.io. Checked 21 October 2026.
  10. The Waymo World Model: A New Frontier for Autonomous Driving Simulation waymo.com. Checked 21 October 2026.
  11. Waymo Research waymo.com. Checked 21 October 2026.