Wenlei Bao
Bellevue, Washington, United States
Summary
🤩
Rockstar🎓
Top SchoolWenlei Bao is a Member of Technical Staff and ML systems engineer with seven years of experience building high-performance compiler and inference infrastructure for leading AI companies. He has driven compiler and on-device ML work at Apple and Microsoft, contributed performance-critical optimizations to the widely used microsoft/onnxruntime (notably AVX/AVX2/AVX512 tensorization for 8/16-bit GEMM/GEMV), and advanced LLM training and inference tooling at ByteDance. Now at Anthropic, he focuses on productionizing scalable ML infra informed by deep academic training (PhD, OSU) and GPU/CNN optimization experience from NVIDIA. Known for squeezing performance from hardware via cross-compilation and parallel tensorization strategies, he pairs low-level systems expertise with practical experience across research and product teams.
7 years of coding experience
11 years of employment as a software developer
Bachelor of Science (BS) & Master of Science (MS) Electrical and Electronics Engineering, Bachelor of Science (BS) & Master of Science (MS) Electrical and Electronics Engineering at Harbin Institute of TechnologyDoctor of Philosophy (PhD) Computer Science and Engineering, Doctor of Philosophy (PhD) Computer Science and Engineering at The Ohio State University
Github Skills (10)
avx10
machine-learning10
deeplearning-ai10
c-language10
deep-learning10
onnx10
cprogramming-language10
hardware-acceleration10
pytorch8
tensorflow8
Programming languages (2)
Github contributions (5)
microsoft/onnxruntime
Sep 2019 - Dec 2019
ONNX Runtime: cross-platform, high performance ML inferencing and training accelerator
Role in this project:
ML Engineer
ML EngineerContributions:11 commits, 20 PRs, 41 pushes in 2 months
Contributions summary:Wenlei primarily focused on optimizing the ONNX Runtime for x86 architectures, specifically for integer matrix multiplication (GEMM) and general matrix-vector multiplication (GEMV) operations. Their contributions included implementing and refining tensorization strategies for 8-bit and 16-bit integer data types, targeting AVX, AVX2, and AVX512 instruction sets. They also addressed cross-compilation issues and integrated parallel execution capabilities within the tensorization framework to improve performance.
runtimetrainingtensorflowai-frameworkaccelerator
baowenlei/tvm
Apr 2019 - Apr 2019
Open deep learning compiler stack for cpu, gpu and specialized accelerators
Contributions:3 pushes, 3 branches in 15 days
cpugpu-programminggpu-accelerationtvmdeep-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.