Yinghai Lu
Member Of Technical Staff at Thinking Machines Lab
Arica, Arica and Parinacota Region, United States
Summary
🤩
Rockstar🎓
Top SchoolYinghai Lu is a seasoned software and systems engineer with over a decade building high-performance inference platforms for top AI labs, most recently leading Research Inference at OpenAI and prior AI inference efforts at Meta. He combines deep research roots—a Ph.D. from Fudan and postdoc experience at Northwestern—with hands-on production expertise in compiler backends, TensorRT, Glow, and PyTorch to deploy generative and recommendation models at scale. Yinghai has repeatedly founded and grown teams (including an accelerator enablement org at Meta and a founding role at Thinking Machines Lab) to bridge hardware, compiler, and system-level optimization. His open-source contributions to prominent projects like Glow, ONNX, and PyTorch demonstrate a knack for low-level operator support and memory/shape debugging that materially improve deployment robustness. Colleagues describe him as a tech lead who moves seamlessly between algorithmic optimization and large-scale distributed engineering. An uncommon strength is his track record of shipping both research-grade inference for cutting-edge models (GPT-4.5, o-series) and pragmatic build/CI improvements that prevent production outages.
10 years of coding experience
14 years of employment as a software developer
B.E. Electrical Engineering, B.E. Electrical Engineering at Tongji UniversityPh.D. Electrical Engineering, Ph.D. Electrical Engineering at Fudan University
English, Chinese
Stackoverflow
315reputation
4kreached
0answers
3questions
Github Skills (42)
pytorch10
c-language10
caffe10
python10
compiler-development10
tensorrt10
glow10
machine-learning10
build-system10
cmake10
onnx10
transformer-models10
nvidia10
deep-learning10
build-automation10
Programming languages (8)
Github contributions (5)
pytorch/glow
Aug 2018 - Aug 2021
Compiler for Neural Network hardware accelerators
Role in this project:
Back-end Developer & ML Engineer
Back-end Developer & ML EngineerContributions:5 reviews, 139 commits, 169 PRs in 3 years 1 month
Contributions summary:Yinghai's contributions primarily focused on enhancing the ONNX importer for the Glow compiler, specifically by integrating support for various operators and fixing bugs. They added support for features like handling specific tensor data types (INT32) and implemented the import of new ONNX operations like 'HalfToFloat', 'Sqr', 'ReduceBackMean', and 'BatchMatMul', indicating expertise in compiler development. Furthermore, the user addressed memory allocation issues and implemented improvements to existing functionality within the ONNX importer, reflecting a deep understanding of the Glow compiler's internals.
hardware-acceleratorscompilerneural-networkacceleratorshardware
pytorch/pytorch
Feb 2018 - Jan 2023
Tensors and Dynamic neural networks in Python with strong GPU acceleration
Role in this project:Backend Developer
Backend DeveloperContributions:22 reviews, 260 commits, 319 PRs in 4 years 11 months
Contributions summary:Yinghai primarily contributed to bug fixes and code improvements within the PyTorch framework. Their work included resolving issues related to `getattr_recursive`, correcting typos in the code, and addressing compilation errors related to `Tensor` ambiguity. They also worked on improving CUDA device error messages and backing out specific changes related to memory efficient attention. These contributions indicate a focus on core functionality and optimization of the library.
pythongpu-accelerationdeep-learninggpunumpy
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.