Liwei Song
Algorithm Scientist at Huawei
Shenzhen, Guangdong Province, China
Summary
👤
Senior🎓
Top SchoolLiwei Song is an Algorithm Scientist at Huawei Cloud with nine years of experience bridging cutting-edge research and production systems in resource optimization for cloud networks. He earned a PhD in Electrical Engineering from Princeton studying security and privacy in machine learning and applied that expertise during an internship at Facebook quantifying practical privacy leakage. At Huawei he focuses on network capacity planning and traffic engineering while also contributing to privacy-preserving ML tooling — notably adding an entropy-based membership inference attack to the TensorFlow Privacy project. Based in Shenzhen, he combines rigorous academic training with hands-on engineering to tackle scalable, security-sensitive problems in cloud infrastructure.
9 years of coding experience
Bachelor’s Degree, Electrical Engineering, Bachelor’s Degree, Electrical Engineering at Peking UniversityDoctor of Philosophy (Ph.D.), Electrical Engineering, 3.966, Doctor of Philosophy (Ph.D.), Electrical Engineering, 3.966 at Princeton University
Chinese, English
Github Skills (7)
data-structures10
machine-learning10
privacy10
tensorflow10
python10
data-structure10
testing9
Programming languages (2)
Github contributions (5)
tensorflow/privacy
Aug 2020 - Dec 2020
Library for training machine learning models with privacy for training data
Role in this project:
ML Engineer
ML EngineerContributions:23 reviews, 30 commits, 3 PRs in 3 months
Contributions summary:Liwei implemented and tested a new membership inference attack technique, specifically adding and testing an entropy-based attack for assessing privacy risks in machine learning models. Their work involved modifying data structures to include entropy calculations and integrating the new attack into the existing membership inference framework. This included adding tests to ensure correct entropy calculations and attack behavior. The changes demonstrate a focus on privacy-preserving machine learning techniques within the context of the TensorFlow Privacy library.
machine-learning-trainingprivacydifferential-privacymachine-learningtraining
inspire-group/privacy-vs-robustness
Aug 2019 - Nov 2019
Contributions:55 commits, 31 pushes, 1 branch in 3 months
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.