Hanxiao Liu
Member Of Technical Staff at Microsoft AI
Mountain View, California, United States
Summary
🤩
Rockstar🎓
Top SchoolHanxiao Liu is a machine learning engineer and researcher with 11 years of experience bridging cutting-edge research and production at top AI organizations, currently a Member of Technical Staff at Microsoft AI. He held research roles at Google Brain and Inflection AI and interned at DeepMind and Citadel, reflecting deep expertise in neural architecture, model evaluation, and scalable ML systems. His open-source contributions include work on the influential DARTS differentiable architecture search repo, adding visualization, testing, and ImageNet evaluation scripts that improved reproducibility and benchmarking. Trained at Carnegie Mellon (PhD) and Tsinghua, he combines rigorous academic foundations with hands-on implementation skills and a knack for turning research prototypes into robust experiments. Colleagues describe him as both detail-oriented in optimization and pragmatic about bringing models to real-world evaluation.
11 years of coding experience
5 years of employment as a software developer
Bachelor's Degree Automation, Bachelor's Degree Automation at Tsinghua UniversityDoctor of Philosophy (PhD) Computer Science, Doctor of Philosophy (PhD) Computer Science at Carnegie Mellon University
Github Skills (11)
mask-rcnn10
faster-rcnn10
rnn-model10
deep-learning10
n10
neural-architecture-search10
image-classification9
pytorch9
automl8
python7
imagenet7
Programming languages (1)
Github contributions (5)
quark0/darts
Jun 2018 - Oct 2018
Differentiable architecture search for convolutional and recurrent networks
Role in this project:
ML Engineer
ML EngineerContributions:67 commits, 1 PR, 59 pushes in 3 months
Contributions summary:Hanxiao contributed to the project by implementing and modifying scripts related to training and testing neural architecture search models. They added and modified visualization scripts for CNN and RNN architectures, enabling the plotting of network structures. Moreover, they added testing scripts, including ones for ImageNet, and fixed training parameters for ImageNet experiments, indicating a focus on model evaluation and optimization.
pytorchdifferentiabledeep-learningconvolutionalmachine-learning
quark0/CGL
Oct 2015 - Aug 2018
Contributions:19 commits, 17 pushes, 1 branch in 2 years 10 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.