Kevin Wu
Senior Software Engineer at Meta
Seattle, Washington, United States
Summary
🤩
Rockstar🎓
Top SchoolKevin Wu is a Senior Software Engineer at Meta with nine years of experience building back-end and ML systems, currently contributing to Meta's GenAI efforts from Seattle. He combines an ECE and Computer Science background from Rice University with hands-on expertise in training and evaluation tooling for large language models, adding features like max_step control, flop counting, and test fixes to the widely used meta-llama/llama-cookbook. At Meta he has progressed from early engineering roles to senior status, focusing on performance measurement and reliable training workflows that help scale fine-tuning efforts. Known for pragmatic engineering that bridges research and production, he improves developer productivity by hardening tests and instrumenting throughput—work that’s especially valuable in high-throughput GenAI pipelines.
9 years of coding experience
5 years of employment as a software developer
Bachelor of Science - BS, Electrical and Computer Engineering, Bachelor of Science - BS, Electrical and Computer Engineering at Rice UniversitySt. Mark's School
Github Skills (9)
pytorch10
machine-learning10
python10
ai10
fine-tuning10
evaluation9
eval9
test-automation8
llm8
Programming languages (3)
Github contributions (5)
meta-llama/llama-cookbook
Apr 2024 - Mar 2025
Welcome to the Llama Cookbook! This is your go to guide for Building with Llama: Getting started with Inference, Fine-Tuning, RAG. We also show you how to solve end to end problems using Llama model family and using them on various provider services
Role in this project:
Back-end & ML Engineer
Back-end & ML EngineerContributions:47 reviews, 30 PRs, 119 pushes in 10 months
Contributions summary:Kevin enhanced the training and evaluation process by implementing a `max_step` feature in the `train_utils.py` file, which provides control over the maximum number of training steps. Furthermore, they addressed issues in the unit tests, specifically fixing the `test_gradient_accumulation` and `test_save_to_json` tests to improve the testing framework's reliability. In addition, the user added a flop counter feature and incorporated it into the training framework to measure model throughput during fine-tuning, aiding in performance analysis.
aifinetuninglangchainllamallama2
wukaixingxp/vllm
Apr 2024 - Oct 2024
Contributions:6 pushes, 1 branch in 5 months
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.