Shuqin Xie

I am a Member of Technical Staff at Amazon AGI SF Lab, where I work on post-training and reinforcement learning for browser-use agents and multimodal productivity. Previously, I was a Senior Applied Scientist and Tech Lead for Scene Understanding at Cruise, developing multimodal perception systems for autonomous driving.

Earlier, I worked on self-driving research at Uber ATG with Prof. Raquel Urtasun and on computer vision research at Shanghai Jiao Tong University with Prof. Cewu Lu. My interests span machine learning, computer vision, robotics, and AI agents.

Email · LinkedIn · Résumé (PDF)

Experience

• Amazon AGI SF Lab San Francisco, California
  Member of Technical Staff, Applied Science September 2025 - Present

  Leading reinforcement-learning and post-training work for browser-use agents; contributed to Amazon Nova Act and multimodal document-understanding systems.

• Cruise LLC San Francisco, California
  Senior Applied Scientist, Tech Lead - Scene Understanding February 2022 - September 2025

  Led the transformation of construction-zone perception from heuristics to a transformer-based, multimodal, multi-task system using semantic-map context and semi-supervised learning.

• Uber Advanced Technologies Group Toronto, Canada
  Research Intern · Advisor: Prof. Raquel Urtasun July 2019 - June 2020

  Developed a prototype-based open-set instance-segmentation approach using dense pixel embeddings, metric learning, and clustering-based inference.

• Machine Vision and Intelligence Group at SJTU Shanghai, China
  Research Assistant · Advisor: Prof. Cewu Lu August 2016 - January 2018

  Researched robust multi-person pose estimation and reinforcement-learning methods for refining imperfect bounding boxes in downstream vision tasks.

Education

• Carnegie Mellon University Pittsburgh, Pennsylvania
  Master of Science in Computer Vision, GPA: 4.11/4.33 August 2020 - December 2021
• Shanghai Jiao Tong University Shanghai, China
  Bachelor of Engineering in Automation, GPA: 3.9/4.3 September 2014 - June 2019

Publications

• Environment Upgrade Reinforcement Learning for Non-differentiable Multi-stage Pipelines
  Shuqin Xie, Zitian Chen, Chao Xu, Cewu Lu
  IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2018 (Spotlight)
• RMPE: Regional Multi-Person Pose Estimation
  Hao-Shu Fang, Shuqin Xie, Yu-Wing Tai, Cewu Lu
  International Conference on Computer Vision (ICCV), 2017

Technical Reports

• TOPNet: Thinking Outside the Bounding Box
  Shuqin Xie, Chao Xu, Shu Liu, Alan Yuille, Jiaya Jia, 2019
• Post-NMS Training Strategy for Object Detection
  Lu Qi*, Shuqin Xie*, Shu Liu, Jiaya Jia, Yue Zhou, 2019