Lei Zhang
高级架构师 at The Apache Software Foundation
Chaoyang District, Beijing, China
Summary
🤩
Rockstar🎓
Top SchoolLei Zhang is a research scientist with 13 years of experience specializing in performance, reliability, and observability for large-scale cloud and microservice systems, currently optimizing LLM training and inference infrastructure at ByteDance. His academic background includes a Princeton postdoc and a PhD from Emory, where he applied static analysis and distributed tracing to automate troubleshooting in distributed systems and built practical volumetric streaming systems. Lei’s work spans distributed cache/memory management, RDMA-based AI infrastructure, and memory-sensitive performance analysis—skills honed during a PhD internship at Facebook on video cache eviction. An active open-source contributor and ASF committer, he contributes backend and DevOps improvements to the popular CAMEL multi-agent AI framework, focusing on data pipelines, configuration robustness, and observability. He combines rigorous research methods with production-facing engineering, often surfacing subtle causes of at-scale performance anomalies that are easy to miss in conventional monitoring.
13 years of coding experience
学士, 电子技术, 学士, 电子技术 at 河北师范大学
Github Skills (12)
multiagent-systems10
multi-agent10
agent10
large-language-models10
python10
natural-language-processing10
data-engineering10
deep-learning8
docker4
dockers4
kubernetes3
kubernetes-pods3
Programming languages (14)
Github contributions (5)
camel-ai/camel
Aug 2024 - Apr 2025
🐫 CAMEL: Finding the Scaling Law of Agents. The first and the best multi-agent framework. https://www.camel-ai.org
Role in this project:
Back-end Developer & DevOps Engineer
Back-end Developer & DevOps EngineerContributions:6 reviews, 23 PRs, 22 pushes in 7 months
Contributions summary:Lei contributed to the CAMEL project by addressing various issues related to model configuration, URL handling, and data generation. They fixed the priority issue in the model configuration for user_agent_kwargs and assistant_agent_kwargs. The user also implemented enhancements for JSON output and added features like the stop_on_first_failure option to instruction filters, demonstrating a focus on refining the data generation pipeline. Furthermore, they improved logging capabilities for error handling and instruction generation progress.
agentai-societiesartificial-intelligencecommunicative-aicooperative-ai
coolbeevip/license-maven-plugin
Jun 2020 - Apr 2024
Contributions:13 releases, 9 PRs, 82 pushes in 3 years 10 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.