Cheng Chen
Software Engineer at Microsoft
Beijing, China
Summary
🤩
Rockstar🎓
Top SchoolCheng Chen is a software engineer specializing in NLP and AI with 11 years of industry experience and a Master’s in Cyber Science from Beihang University. He has applied ML and deep learning to real-world products across Sohu, Shopee, Kuaishou, DiDi, Baidu and now Microsoft, focusing on news recommendation, product matching, and automated video/text generation. At Sohu he optimized semantic vectors and topic features to boost recommendation performance, and at Kuaishou his summarization work helped generate videos that reached millions of views. Cheng also contributes to open-source ML tooling, extending Microsoft’s MMdnn to support additional PyTorch operators and conversion paths—demonstrating practical cross-framework expertise. Based in Beijing, he combines research-rooted techniques with production engineering to turn NLP models into measurable product gains and is actively seeking impactful collaborations.
10 years of coding experience
4 years of employment as a software developer
Bachelor's degree, Information Security, Bachelor's degree, Information Security at Yunnan UniversityMaster's degree, Cyber Science and Technology, Master's degree, Cyber Science and Technology at Beihang University
Github Skills (8)
keras10
pytorch10
machine-learning10
model-conversion10
deep-learning10
tensorflow10
neural-network9
convolutional-neural-networks9
Programming languages (3)
Github contributions (5)
microsoft/MMdnn
Nov 2017 - Oct 2018
MMdnn is a set of tools to help users inter-operate among different deep learning frameworks. E.g. model conversion and visualization. Convert models between Caffe, Keras, MXNet, Tensorflow, CNTK, PyTorch Onnx and CoreML.
Role in this project:
ML Engineer
ML EngineerContributions:7 releases, 456 commits, 206 PRs in 11 months
Contributions summary:Cheng's commits focus on modifying and extending the codebase to support the integration and conversion of machine learning models. The code changes involve adding support for new layer types within the conversion pipeline. Specifically, the user contributes to the implementation of the PyTorch emitter and includes modifications to enable support for operators like ReLU6, crop and depthwise convolution, and also added test examples.
caffe2intertensorflowmodel-conversionoperate
kitstar/DNNConvert
May 2017 - Aug 2017
Contributions:38 commits, 33 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.