Xiao-yong Jin
Assistant Computational Scientist
Lemont, Illinois, United States
Summary
👤
Senior🎓
Top SchoolXiao-yong Jin is an Assistant Computational Scientist at Argonne National Laboratory with 11 years of experience applying high-performance computing to fundamental physics problems like lattice QCD and lattice gauge theories. He holds a Ph.D. in Physics from Columbia and blends deep theoretical knowledge with practical systems engineering, demonstrated by contributions to HPC-focused open-source projects. His work spans low-level system fixes for plan9port, package manager improvements for Nimble, and advanced ML engineering in llama.cpp where he implemented customizable RoPE to extend context windows. At Argonne he moved from postdoctoral research to a staff scientist role, translating Monte Carlo and lattice techniques into production-grade code on supercomputers. Comfortable across C/C++ and system tooling, he brings a rare combination of HEP theory expertise and hands-on software craftsmanship that improves both scientific models and the tools that run them.
11 years of coding experience
6 years of employment as a software developer
Bachelor of Science (B.S.), Physics, Bachelor of Science (B.S.), Physics at Fudan UniversityPh.D., Physics, Ph.D., Physics at Columbia University in the City of New York
Chinese, Japanese, English
Github Skills (20)
nim-lang10
c-language10
system-programming10
back-end-development10
llama10
memory-management10
c1110
llm-inference10
c1710
package-manager-tool10
gm10
macos10
package-manager10
cprogramming-language10
filesystem9
Programming languages (18)
Github contributions (5)
9fans/plan9port
Mar 2018 - Jan 2021
Plan 9 from User Space
Role in this project:
Back-end Developer & System Engineer
Back-end Developer & System EngineerContributions:1 review, 31 commits, 26 PRs in 2 years 11 months
Contributions summary:Xiao-yong primarily contributed to the Plan 9 from User Space project by addressing memory leaks and fixing bugs related to file system interaction and font rendering. They implemented fixes for macOS-specific issues, including improving the handling of pseudoterminals and enhancing support for HiDPI displays. Their contributions involved modifying core system components and addressing platform-specific compatibility issues.
user-spaceplan
ggml-org/llama.cpp
Mar 2023 - Sep 2024
LLM inference in C/C++
Role in this project:ML Engineer
ML EngineerContributions:11 reviews, 7 PRs, 101 comments in 1 year 6 months
Contributions summary:Xiao-yong contributed to the implementation of customizable RoPE (Rotary Positional Embedding) within the `llama.cpp` codebase. This involved adding new parameters for frequency base and scale, directly impacting context window extension. The user's work includes modifying the core `llama.cpp` and `ggml.c` files, as well as integrating these changes across the model loading and evaluation processes, demonstrating a deep understanding of the underlying mechanisms of the model. Additionally, the user's commits included the implementation of features related to the server component of the project, specifically allowing for the input of token arrays, and the addition of the option to disable KV offload.
ggmlllama
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.