Ke Ding
Principal Algorithm Expert at Meituan
Beijing, China
Summary
👤
Senior🎓
Top SchoolKe Ding is a Principal Algorithm Expert based in Beijing with 11 years designing and deploying large-scale speech recognition and deep learning systems across top-tier companies including Meituan, Tencent, Baidu and Microsoft Research Asia. He holds a PhD in Computer Science from Peking University and has led production HF/ASR work—far-field ASR, distributed training (Kaldi Nnet3, PyTorch), dialogue systems and speech-driven talking-face generation with GANs. His background spans HPC for GPU clusters, acoustic modeling (CNN/LSTM/CTC/attention), speaker/dialect recognition and model compression including binary/ternary networks. An active open-source contributor, he’s fixed bugs and added features to the widely used Keras library and published experimental Keras implementations exploring XNOR/ternary layers and novel activations like SReLU. Known for bridging research and engineering, he combines deep theoretical knowledge with hands-on systems optimization to ship robust production models at scale. A less obvious strength is his repeated work on low-precision and sparse model techniques that enable efficient deployment in constrained environments.
11 years of coding experience
3 years of employment as a software developer
Doctor of Philosophy (PhD), Computer Science, Doctor of Philosophy (PhD), Computer Science at Peking UniversityBachelor of Engineering (B.E.), Computer Science, Bachelor of Engineering (B.E.), Computer Science at China Agricultural University
English
Stackoverflow
1reputation
0reached
0answers
0questions
Github Skills (16)
neural-network10
keras10
machine-learning10
recurrent-neural-networks10
deep-learning10
tensorflow10
python10
data-science9
faster-rcnn8
mask-rcnn8
data-structure7
data-structures7
algorithm7
algorithms7
pytorch6
Programming languages (4)
Github contributions (5)
DingKe/nn_playground
Dec 2016 - Sep 2018
Experimental keras implementation of novel neural network structures
Role in this project:
ML Engineer
ML EngineerContributions:1 release, 28 commits, 4 PRs in 1 year 9 months
Contributions summary:Ke primarily contributed to the implementation of novel neural network structures within the experimental Keras framework. Their work involved updating the codebase to Keras 2.0, integrating new layers like those for XNOR and ternary networks, and experimenting with weight normalization. The commits showcase modifications to existing network architectures, suggesting a focus on exploring and integrating different model types and optimization techniques.
deep-learningneural-networksmachine-learningneural-networktensorflow
keras-team/keras
Jan 2016 - Dec 2016
Deep Learning for humans
Role in this project:ML Engineer
ML EngineerContributions:7 commits, 15 PRs, 25 comments in 11 months
Contributions summary:Ke contributed to the Keras deep learning library by fixing bugs and adding new functionalities. Their work includes fixing an issue related to stateful RNNs, adding support for saving and loading sample weight modes in models, and making the sparse crossentropy function more robust. The user also added a new activation function, SReLU, demonstrating a focus on improving and extending the core functionalities of the library.
pythondata-sciencedeep-learningneural-networksmachine-learning
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.