Yiheng Xu
Software Engineer at DoorDash
Denver, Colorado, United States
Summary
🤩
Rockstar🎓
Top SchoolYiheng Xu is a software engineer with nine years of experience building reliable, production-scale systems in cloud and HPC environments, currently at DoorDash after contributing core services at Palantir. He designs developer-friendly SDKs and distributed compute platforms (an AWS Lambda–like service) with a focus on automation, observability, RBAC, and data lineage for secure, auditable execution. Past research and internships at Lawrence Livermore and in academia reflect deep HPC and I/O performance expertise, including tools and ML models to characterize burst-buffer suitability and enable multi-tool profiling. He also contributes to prominent open-source ML work—notably on Microsoft’s LayoutLM models for document understanding—bringing applied ML engineering experience to system-level problems. Based in Denver, he combines production systems craftsmanship with a research-driven curiosity that surfaces in cross-domain sync and performance-optimization work.
9 years of coding experience
4 years of employment as a software developer
Associate of Science - AS Computer Science, Associate of Science - AS Computer Science at Wuhan UniversityBachelor of Science - BS Computer Science, Bachelor of Science - BS Computer Science at University of Wisconsin-MadisonMaster of Science - MS Computer Science, Master of Science - MS Computer Science at University of MarylandComputer Science, Computer Science at UC Berkeley Extension
English, Chinese
Github Skills (7)
transformer-models10
pytorch10
machine-learning10
nlp10
fine-tuning9
python9
computer-vision9
Programming languages (4)
Github contributions (5)
microsoft/unilm
Feb 2020 - Nov 2021
Large-scale Self-supervised Pre-training Across Tasks, Languages, and Modalities
Role in this project:
ML Engineer
ML EngineerContributions:3 reviews, 31 commits, 15 PRs in 1 year 9 months
Contributions summary:Yiheng's contributions primarily revolve around the "LayoutLM" and "LayoutLMv2" models within the `microsoft/unilm` repository, indicating a focus on document understanding and information extraction. The commits demonstrate efforts to release these models, fix related issues, and handle exceptions within the document processing pipeline. The user also worked on updating the code, especially around model configurations, and CUDA device handling.
layoutxlmtraininglanguage-understandingvision-and-languagewavlm
ranpox/RSSHub
May 2018 - Jan 2025
🍰 Everything is RSSible
Contributions:25 PRs, 63 pushes, 7 branches in 6 years 9 months
everythingrssrss-reader
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.