Juechu Dong
Deep Learning Architect Intern at University of Michigan
Ann Arbor, Michigan, United States
Summary
🤩
Rockstar🎓
Top SchoolJuechu (Joy) Dong is a PhD candidate at the University of Michigan and a Deep Learning Architect Intern at NVIDIA, specializing in GPU architecture, confidential computing, and privacy-preserving analytics. With seven years of experience spanning research internships at Meta/PyTorch and production-focused GPU performance work at NVIDIA, she bridges systems research and practical deep learning engineering. Her contributions to PyTorch’s flex_attention—improving LLM inference, testing, and benchmarking—underscore a rare blend of kernel-level optimization and ML systems know-how. Joy’s research targets confidential computing for large-scale genomic analysis and generative AI, pairing Trusted Execution Environment expertise with GPU kernel tuning. Known for academic rigor (full scholarship-level performance in dual bachelor programs) and hands-on impact, she thrives at the intersection of privacy, performance, and scalable ML infrastructure.
7 years of coding experience
1 year of employment as a software developer
Doctor of Philosophy - PhD, Computer Science and Engineering, Doctor of Philosophy - PhD, Computer Science and Engineering at University of MichiganBachelor's degree, Electrical and Computer Engineering, Bachelor's degree, Electrical and Computer Engineering at UM-SJTU Joint Institute, Shanghai Jiao Tong UniversityBachelor's degree, Electrical and Computer Eningeering, 3.82/4.00, Bachelor's degree, Electrical and Computer Eningeering, 3.82/4.00 at Shanghai Jiao Tong University
Github Skills (9)
pytorch10
machine-learning10
tensor10
deep-learning10
gpu9
triton9
neural-network8
autograd8
python8
Programming languages (6)
Github contributions (5)
pytorch/pytorch
May 2024 - Dec 2024
Tensors and Dynamic neural networks in Python with strong GPU acceleration
Role in this project:
ML Engineer
ML EngineerContributions:43 reviews, 33 PRs, 233 pushes in 7 months
Contributions summary:Juechu contributed to the development and testing of the `flex_attention` kernel, a higher-order operation within the PyTorch framework designed for efficient attention mechanisms. Their contributions included implementing strided input tests, enhancing flex decoding capabilities for LLM inference, and adding a benchmark for flex decoding performance. They also added support for explicit GQA and refined kernel parameters for improved performance.
pythongpu-accelerationdeep-learninggpunumpy
joydddd/joydddd.github.io
Mar 2021 - Mar 2025
Contributions:38 pushes, 3 branches in 4 years
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.