Chang Lan
Member Of Technical Staff at Anthropic
San Francisco Bay Area United States
Summary
🤩
Rockstar🎓
Top SchoolChang Lan is a Member of Technical Staff in the Bay Area with 12 years of experience building high-performance distributed systems and ML infrastructure across top AI labs and cloud providers. He has driven foundation-model pretraining and serving platforms at Apple, DeepMind, and now Anthropic, blending research insight from a PhD at UC Berkeley with production-grade engineering. His open-source contributions span low-level networking and distributed training projects—RDMA improvements in ps-lite, refactors in Berkeley’s BESS switch, and MLOps automation for Vertex AI and Horovod—demonstrating a focus on reliability and performance. He’s comfortable across the stack from CUDA and memory-pool optimizations to orchestration and deployment automation, and has hands-on experience shaping build and packaging workflows. Colleagues rely on him to untangle complex concurrency and communication issues, and he brings a quieter but persistent bias for refactoring and codebase hygiene that often unlocks larger performance gains.
12 years of coding experience
9 years of employment as a software developer
Doctor of Philosophy - PhD, Computer Science, Doctor of Philosophy - PhD, Computer Science at University of California, BerkeleyBachelor's degree, Computer Science, Bachelor's degree, Computer Science at Tsinghua University
English, Chinese
Stackoverflow
313reputation
7kreached
2answers
3questions
Github Skills (48)
docker10
c-language10
distributed-training10
rdma10
python10
memory-management10
machine-learning10
horovod10
networking10
dockers10
gcp10
tensorflow10
vertex-ai10
cprogramming-language10
architecture9
Programming languages (14)
Github contributions (5)
NetSys/bess
Nov 2016 - Oct 2017
BESS: Berkeley Extensible Software Switch
Role in this project:
Back-end Developer
Back-end DeveloperContributions:5 releases, 47 commits, 180 PRs in 10 months
Contributions summary:Chang's contributions primarily involved cleaning up and refactoring the codebase, specifically focusing on removing unused components and renaming functions and types related to a "snobj" feature. This included significant code modifications such as renaming, moving, and removing code related to initialization and command functions. Furthermore, the user updated documentation, fixed indentation, and made improvements to ensure command thread safety.
berkeleybessswitchextensible
dmlc/ps-lite
Jul 2018 - Aug 2019
A lightweight parameter server interface
Role in this project:Back-end Developer
Back-end DeveloperContributions:51 commits, 1 PR, 5 comments in 1 year 1 month
Contributions summary:Chang primarily contributed to the RDMA (Remote Direct Memory Access) implementation within the ps-lite project. Their work involved refactoring the RDMA-based communication layer, including memory management with a custom memory pool, and optimizing message handling. The user also addressed connection issues and enhanced the reliability and efficiency of the RDMA communication protocol within the parameter server framework.
parameter-serverparameter
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.