Jiachen Xu is a Senior Machine Learning Engineer based in Beijing with 11 years of experience building production ML systems across tech and finance, including ByteDance, Amazon, S&P Global, and J.P. Morgan. He specializes in recommendation and social feed models, having designed a Trilateral Share model and end-to-end repost distribution pipeline that materially increased engagement and DAU through techniques like hard negative resampling and FocalLoss. Jiachen blends strong academic foundations—a Columbia MS in Data Science and Stanford AI certificate—with hands-on engineering, shipping robust features and distribution logic that lift key business metrics. His background spans research-grade probabilistic models (Mixture Markov Models, knockoffs) to practical improvements in market-impact neural predictors, demonstrating both theoretical depth and applied impact. He also contributes to open-source front-end tooling, improving HLS streaming robustness in ByteDance’s xgplayer, signaling a willingness to work across the stack. Colleagues describe him as a metrics-driven engineer who pairs experimental rigor with production-minded implementation.
11 years of coding experience
4 years of employment as a software developer
Master's Degree, Data Science Instritute, 3.92/4.0, Master's Degree, Data Science Instritute, 3.92/4.0 at Columbia University in New York City
Bachelor's degree, Applied Mathematics, Major GPA: 3.88/4.0, Bachelor's degree, Applied Mathematics, Major GPA: 3.88/4.0 at Shanghai Jiao Tong University
University College London
Artificial Intelligence Graduate Certificate, Artificial Intelligence, 4.0, Artificial Intelligence Graduate Certificate, Artificial Intelligence, 4.0 at Stanford University
A HTML5 video player with a parser that saves traffic
Role in this project:
Front-end Developer
Contributions:187 reviews, 21 commits, 160 PRs in 6 months
Contributions summary:Jiachen focused on enhancing the functionality and stability of the HTML5 video player, specifically addressing issues related to HLS streaming. They fixed HLS.js-related problems, such as ensuring proper destruction and handling of dynamic URL updates. The user also contributed to improving the player's overall robustness by synchronizing the destroy API's behavior. Furthermore, they implemented a feature to dynamically update the HLS URL, showcasing their involvement in improving the player's streaming capabilities.
Swfpack is a tool that convert swf file format between uncompressed/zlib/lzma.
Contributions:3 PRs, 26 pushes, 1 branch in 8 years 4 months
uncompressedfile-formatcompressionzlibswf
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.
Request Free Trial
Jiachen Xu - Senior Machine Learning Engineer at ByteDance