Jiaming Yuan
Engineer at NVIDIA
Guangzhou City, Guangdong Province, China
Summary
🤩
RockstarJiaming Yuan is an experienced engineer based in Guangzhou with nine years specializing in high-performance ML systems and backend development. Currently at NVIDIA, he has substantial open-source impact, contributing to flagship projects like XGBoost and Optuna by modernizing pruning callbacks, quantile objectives, and categorical feature handling. His work on RAPIDS (cuML and raft) shows deep expertise in CUDA-accelerated algorithms, deterministic numerical behavior, and mdspan/device integration for efficient matrix operations. Comfortable across C++, Python, and distributed setups (Dask/external memory), he blends low-level optimization with production-ready testing and build improvements. Colleagues can expect a pragmatic engineer who reduces complexity—fixing memory leaks, build issues, and test flakiness—while advancing ML tooling used broadly in the community.
9 years of coding experience
Github Skills (29)
xgboost10
c-language10
python10
gpu-programming10
categorical-data10
machine-learning10
machine-learning-algorithms10
regression-analysis10
dm10
kernel10
hyperparameter-optimization10
cuda10
logging10
cuml10
cprogramming-language10
Programming languages (16)
Github contributions (5)
dmlc/xgboost
May 2018 - Jan 2023
Scalable, Portable and Distributed Gradient Boosting (GBDT, GBRT or GBM) Library, for Python, R, Java, Scala, C++ and more. Runs on single machine, Hadoop, Spark, Dask, Flink and DataFlow
Role in this project:
Back-end Developer & ML Engineer
Back-end Developer & ML EngineerContributions:25 releases, 2522 reviews, 1213 commits in 4 years 8 months
Contributions summary:Jiaming's contributions primarily revolved around enhancing the XGBoost library, specifically focusing on implementing tests for quantile-based objectives and improving the handling of categorical features. The user added support for multi-target trees and fit intercept for the hinge objective. They also refined the data interface, making updates and cleanups to improve performance and consistency, especially within the context of dask and external memory. The user also helped in creating the building pipeline with the implementation of the correct default config.
xgboostpythonflinkdaskdataflow
dmlc/dmlc-core
Oct 2018 - Jun 2022
A common bricks library for building scalable and portable distributed machine learning.
Role in this project:Back-end Developer
Back-end DeveloperContributions:8 reviews, 17 commits, 20 PRs in 3 years 9 months
Contributions summary:Jiaming contributed to the core functionality and maintenance of the `dmlc-core` library, focusing on improvements to logging, parsing, and file system operations. Their work included enabling runtime stacktrace size configuration, suppressing compiler warnings, and adding an installation target within the build process. The user also addressed memory leaks and removed deprecated parameters, demonstrating a commitment to code quality and efficiency. These changes primarily involved modifications to headers, source files, and test code.
scalablebricksmachine-learningportabledistributed-machine-learning
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.