Xiao Li is a Machine Learning Engineer with 17 years of experience, currently working on post-training LLMs and GenAI agents at Meta after leading large-scale NLP model development and deployment as a Senior Applied Scientist at Microsoft. He combines deep academic training—a PhD in Psychometrics/Statistics and an MS in Applied Mathematics from UIUC—with practical production engineering across ML, back-end APIs, and DevOps. His open-source contributions span computer vision (enhancing descriptor matching in the ruby-opencv gem), blockchain tooling (improving JSON-RPC in Diem), and CI/CD server robustness (GoCD agent and protocol improvements), reflecting a rare blend of research rigor and systems-level pragmatism. Notably, his background in temporal knowledge graphs and reinforcement learning informs his ability to design robust, edge-case-resistant ML services for real-world systems.
17 years of coding experience
6 years of employment as a software developer
Bachelor's degree Mathematics, Bachelor's degree Mathematics at Fudan University
Doctor of Philosophy - PhD Psychometrics-Statistics/Quantitative Methods, Doctor of Philosophy - PhD Psychometrics-Statistics/Quantitative Methods at University of Illinois Urbana-Champaign
Diem’s mission is to build a trusted and innovative financial network that empowers people and businesses around the world.
Role in this project:
Back-end Developer
Contributions:532 reviews, 268 commits, 150 PRs in 1 year 5 months
Contributions summary:Xiao primarily contributed to improving the JSON-RPC interface for the Diem blockchain. Their work focused on refining the API's functionality, including fixing transaction hash issues, renaming methods for clarity, and making versioning optional. They also addressed bug fixes related to mempool, health checks, and added tests to improve code coverage. Further work centered around implementing improvements to handle edge cases related to the API.
Contributions:53 commits, 26 PRs, 88 comments in 3 months
Contributions summary:Xiao primarily contributed to the GoCD server, fixing bugs related to JUnit XML parsing and minified files, and also implemented improvements to agent communication. Their work involved changes to core configuration and API code, alongside the testing infrastructure. Additionally, they focused on performance enhancements related to agent improvements, including websocket server modifications and updates to the agent-server communication protocol.
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.