Jiezhong Qiu
Research Assistant at Knowledge Engineering Group of Tsinghua University
Haidian District, Beijing, China
Summary
🤩
Rockstar🎓
Top SchoolJiezhong Qiu is a research-focused machine learning engineer with 11 years of experience bridging academic research and industry R&D, currently working at Tsinghua's Knowledge Engineering Group. His work spans large-scale data mining on xuetangX to predictive models of student learning effectiveness, and he has a strong track record of research internships at top labs including Facebook AI, Microsoft Research Asia, Tencent, and YouTube. He contributes to open-source ML infrastructure—most notably implementing performance-critical components for a fast Mixture-of-Experts (MoE) in PyTorch, including custom CUDA kernels and memory-to-memory attention—which highlights his systems-aspects expertise in deep learning. Pursuing a PhD in Computer Science at Tsinghua, he pairs rigorous academic training with practical deployment experience across NLP, knowledge graphs, and networking systems. Colleagues would describe him as a pragmatic researcher who moves ideas into efficient code, often optimizing low-level kernels to make advanced models run at scale.
11 years of coding experience
Doctor of Philosophy (PhD) Computer Science, Doctor of Philosophy (PhD) Computer Science at Tsinghua University
English, Chinese
Github Skills (5)
cuda10
pytorch10
machine-learning10
deep-learning10
attention-mechanism9
Programming languages (7)
Github contributions (5)
laekov/fastmoe
Nov 2020 - Nov 2021
A fast MoE impl for PyTorch
Role in this project:
ML Engineer
ML EngineerContributions:12 reviews, 143 commits, 12 PRs in 1 year
Contributions summary:Jiezhong primarily focused on implementing and modifying components for a fast Mixture of Experts (MoE) implementation within a PyTorch environment. Their contributions included adding and refining memory-to-memory attention mechanisms, and incorporating different position-wise feed-forward network configurations, including hierarchical and sparse approaches. They also worked on resolving bugs, and incorporating a customized CUDA kernel for matrix multiplication within the MoE architecture.
pytorchmoemixture-of-expertsimplpytorch-lightning
xptree/NetSMF
Feb 2019 - Sep 2019
NetSMF: Large-Scale Network Embedding as Sparse Matrix Factorization
Contributions:22 commits, 3 PRs, 19 pushes in 6 months
social-network-analysissparse-matrixfactorizationnetwork-embeddingsparse
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.