Ming Sun
Senior Applied Research Scientist at Facebook
Seattle, Washington, United States
Summary
🤩
Rockstar🎓
Top SchoolMing Sun is a Senior Applied Research Scientist in Seattle with nine years of industry experience applying advanced ML and NLP techniques at Facebook and Amazon after a PhD in Language Technologies from Carnegie Mellon. He blends deep research rigor—evident from a CMU PhD and postdoc work on spoken dialog systems—with product-driven applied science, shipping solutions in large-scale production environments. His GitHub contributions include work on influential pruning research (ICLR 2019 rethinking-network-pruning), showing practical expertise in model efficiency and lottery-ticket style pruning methods. Comfortable moving between prototyping and production, he focuses on optimizing model compute and robustness while keeping user-facing constraints in mind. Colleagues describe him as a pragmatic researcher who surfaces subtle algorithmic improvements that yield measurable system-level gains.
10 years of coding experience
4 years of employment as a software developer
Master’s Degree, Language Technologies, Master’s Degree, Language Technologies at Carnegie Mellon UniversityBachelor’s Degree, Bachelor’s Degree at Shanghai Jiao Tong University
English, Chinese
Github Skills (6)
pytorch10
machine-learning10
deep-learning10
model-optimization10
python9
imagenet7
Programming languages (4)
Github contributions (5)
Eric-mingjie/rethinking-network-pruning
Oct 2018 - Apr 2020
Rethinking the Value of Network Pruning (Pytorch) (ICLR 2019)
Role in this project:
ML Engineer
ML EngineerContributions:27 commits, 2 PRs, 26 pushes in 1 year 6 months
Contributions summary:Ming primarily worked on a project focused on network pruning for deep learning models using PyTorch. Their commits involve modifications to existing code for computing FLOPs and fixing minor bugs, and include adding and modifying code related to soft pruning and lottery tickets for weight pruning. These changes suggest the user is involved in experimenting with different pruning techniques and optimizing model efficiency within the context of the project.
pytorchconvolutional-neural-networksnetwork-pruningdeep-learning
locuslab/breaking-poisoned-classifier
Oct 2020 - Jan 2022
Code for paper "Poisoned classifiers are not only backdoored, they are fundamentally broken"
Contributions:5 commits, 3 pushes, 1 branch in 1 year 2 months
classifiers
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.