Xinlei Chen
Member Of Technical Staff at xAI
United States
Summary
🤩
Rockstar🎓
Top SchoolXinlei Chen is a machine learning researcher and engineer with 12 years of experience spanning academia and industry, currently a Member of Technical Staff at xAI focusing on multimodal pre-training. He earned a PhD in Artificial Intelligence from Carnegie Mellon and spent seven years at Meta FAIR driving foundational research in vision, language, and multimodal models. Xinlei combines deep research instincts from robotics and vision with practical engineering—contributing to widely used object-detection tooling such as TensorFlow Faster R-CNN by improving COCO dataset integration and training pipelines. His background includes internships and research roles at Google and Microsoft, giving him strong production ML and cloud experience alongside academic rigor. Colleagues describe him as someone who bridges novel model ideas and robust dataset engineering, often surfacing subtle data-handling fixes that materially improve model reliability.
12 years of coding experience
15 years of employment as a software developer
Doctor of Philosophy (Ph.D.) Artificial Intelligence, Doctor of Philosophy (Ph.D.) Artificial Intelligence at Carnegie Mellon UniversityBachelor Computer Science, Bachelor Computer Science at Zhejiang University
English, Chinese
Github Skills (8)
object-detection10
computer-vision10
faster-rcnn10
machine-learning10
coco10
mscoco10
tensorflow10
python10
Programming languages (3)
Github contributions (5)
endernewton/tf-faster-rcnn
Jan 2017 - Sep 2021
Tensorflow Faster RCNN for Object Detection
Role in this project:
ML Engineer
ML EngineerContributions:1 release, 196 commits, 29 PRs in 4 years 8 months
Contributions summary:Xinlei primarily contributed to the COCO dataset integration within the Faster R-CNN framework, adding and modifying code related to dataset loading and annotation handling. They implemented functionality for loading annotations, managing crowd instances, and incorporating pre-computed proposals for various proposal methods. Further contributions include bug fixes and adjustments to the model configuration files, ensuring the correct behavior of the training and testing pipelines. These changes were essential to the proper functioning of the object detection model.
object-detectionrcnnfastermask-rcnntensorflow
facebookresearch/grid-feats-vqa
Apr 2020 - Jul 2020
Grid features pre-training code for visual question answering
Contributions:16 commits, 11 pushes, 3 branches in 2 months
questionmachine-learningtrainingpre-trainingvisual-question-answering
Find and Hire Top DevelopersWe’ve analyzed the programming source code of over 60 million software developers on GitHub and scored them by 50,000 skills. Sign-up on Prog,AI to search for software developers.