Kaichen Liu
Senior Software Engineer at Meta
United States
Summary
🤩
Rockstar🎓
Top SchoolKaichen Liu is a Senior Software Engineer with 8 years of experience building on-device computational photography and ML acceleration for wearable camera platforms at Meta Reality Labs. He combines systems-level expertise across Android/Linux camera stacks, Qualcomm ISPs, and performance profiling with a strong track record in test automation and reliability engineering. Kaichen has contributed to high-profile open source projects like PyTorch, improving Vulkan backend performance and adding operators and optimized matrix-multiplication tests that reduce shader latency on Android GPUs. His background spans cloud and datacenter tooling at AWS, low-latency financial services at Bloomberg, and embedded automation—demonstrating versatility from full-stack dashboards to edge ML. A UC Berkeley EECS graduate, he pairs rigorous academic training with hands-on hardware and firmware experience, often focusing on squeezing real-world performance out of constrained devices. Colleagues know him for pragmatic optimization work that bridges research-quality ML kernels and production-grade embedded systems.
8 years of coding experience
4 years of employment as a software developer
Bachelor of Science, Electrical Engineering and Computer Sciences, 3.51/4.00, Bachelor of Science, Electrical Engineering and Computer Sciences, 3.51/4.00 at University of California, Berkeley
Chinese
Github Skills (13)
vulkan10
pytorch10
machine-learning10
tensor10
gpu10
performance-optimization10
glsl9
cprogramming-language9
c-language9
deep-learning9
neural-network8
python7
autograd6
Programming languages (3)
Github contributions (5)
pytorch/pytorch
Aug 2022 - Aug 2022
Tensors and Dynamic neural networks in Python with strong GPU acceleration
Role in this project:
MLOps Engineer
MLOps EngineerContributions:8 reviews, 2 commits, 6 PRs in 1 day
Contributions summary:Kaichen contributed significantly to the PyTorch Vulkan backend, demonstrating a focus on performance optimization and expanding feature support. Their work included adding Vulkan support for `at::select.int` operator, bilinear upsampling, and softmax, extending functionality to 4-dimensional tensors. They also addressed performance bottlenecks by implementing an improved matrix multiplication algorithm to reduce shader latency on Android GPUs and introduced a dedicated matrix multiplication performance test binary.
pythongpu-accelerationdeep-learninggpunumpy
liuk22/pytorch
Aug 2022 - Nov 2023
Tensors and Dynamic neural networks in Python with strong GPU acceleration
Contributions:30 pushes, 7 branches in 1 year 2 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.