Xiangyi Li

Founder at BenchFlow

San Francisco, California, United States
email-iconphone-icongithub-logolinkedin-logotwitter-logostackoverflow-logofacebook-logo
Join Prog.AI to see contacts
email-iconphone-icongithub-logolinkedin-logotwitter-logostackoverflow-logofacebook-logo
Join Prog.AI to see contacts

Summary

🤩
Rockstar
🎓
Top School
Xiangyi Li is a founder and software engineer with seven years of experience building AI evaluation platforms and open-source tooling from San Francisco. He founded BenchFlow, which rapidly grew SkillsBench to a 600+ member community and runs 60+ benchmarks (90k ARR), applying RL-style environments to evaluate agents on high-GDP domains from spreadsheets to wet lab automation. Before that he was the first founding engineer at OmniAI, where he scaled the 11.9k-star zerox OCR library and contributed backend improvements for robust PDF/file conversion and extraction. His internships at Tesla, Dolby, and Red Hat sharpened his skills in model evaluation, real-time media integration, and cloud-native tooling. Xiangyi blends hands-on backend engineering, evaluation design, and community-driven productization—and often chooses building new ecosystems over traditional corporate offers.
code7 years of coding experience
job1 year of employment as a software developer
bookMaster of Science - MS, Computer Science, Master of Science - MS, Computer Science at San José State University
bookBachelor of Science - BS, Computer Science, Bachelor of Science - BS, Computer Science at The Chinese University of Hong Kong, Shenzhen 香港中文大学(深圳)
github-logo-circle

Github Skills (7)

ocr10
pdf10
file-conversion10
ocra10
regex10
python9
openai-api8

Programming languages (13)

C++CSSRustGoJupyter NotebookKotlinTypeScriptShell

Github contributions (5)

github-logo-circle
getomni-ai/zerox

Aug 2024 - Sep 2025

OCR & Document Extraction using vision models
Role in this project:
userBack-end Developer
Contributions:14 reviews, 8 PRs, 23 pushes in 1 year
Contributions summary:Xiangyi primarily focused on enhancing the repository's functionality to support various document formats for OCR and extraction. They implemented and refined file conversion logic, converting files to PDF format when needed. Furthermore, the user updated the code to handle file extensions, including refining regex logic to improve text formatting. In addition, they added model parameters to restrict max tokens.
ocrpdf
benchflow-ai/skillsbench

Dec 2025 - Jul 2026

SkillsBench evaluates how well skills work and how effective agents are at using them.
Contributions:1 release, 103 reviews, 194 PRs in 6 months
benchmarkskillsbench
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