Yanli Zhao
Senior Staff Software Engineer at Meta
Palo Alto, California, United States
Summary
🤩
Rockstar🎓
Top SchoolYanli Zhao is a Senior Staff Software Engineer in Palo Alto with seven years of experience building large-scale distributed systems and AI/ML infrastructure, currently driving recommendation model and infra co-design at Meta. She has deep expertise in distributed training and inference—contributing to PyTorch core components like DDP, FSDP, RPC and improving GPU memory management and profiling tools used by the community. Her background spans ads delivery and measurement systems to large-scale retrieval inference, including SSD offloading and sparse/dense scaling for foundational models, demonstrating a rare blend of low-level performance work and system-level architecture. A fast learner and problem-solver with an academic grounding in geographic information science and electrical engineering, she pairs algorithmic rigor with hands-on contributions to high-profile open source projects.
7 years of coding experience
1 year of employment as a software developer
University of Illinois Urbana-ChampaignBachelor And Master Of Engineering, Electrical Engineering, Bachelor And Master Of Engineering, Electrical Engineering at University of Electronic Science and Technology of China
Github Skills (16)
cuda10
debugging10
memory-management10
pytorch10
debug10
python10
github9
gpu9
cprogramming-language9
machine-learning9
c-language9
deep-learning8
tensor8
deeplearning-ai8
neural-network8
Programming languages (4)
Github contributions (5)
pytorch/pytorch
Jul 2019 - Jan 2023
Tensors and Dynamic neural networks in Python with strong GPU acceleration
Role in this project:
Back-end Developer
Back-end DeveloperContributions:825 reviews, 310 commits, 204 PRs in 3 years 6 months
Contributions summary:Yanli primarily contributed to bug fixes and enhancements within the PyTorch codebase, with a focus on distributed training and memory management. They addressed issues related to DistributedDataParallel (DDP), particularly when handling zero-output features, and improved the robustness of the cublasLtMatmul kernel. Additionally, the user developed and refined a memory-tracking tool to profile operator-level memory usage, including the ability to save and load memory stats for later analysis and plotting. Furthermore, the user made modifications to input handling and FSDP components within the codebase.
pythongpu-accelerationdeep-learninggpunumpy
zhaojuanmao/pytorch
Apr 2019 - Feb 2024
Tensors and Dynamic neural networks in Python with strong GPU acceleration
Contributions:101 pushes, 42 branches in 4 years 10 months
pythongpu-accelerationdeep-learninggpuacceleration
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.