Xinran Xu
Technical Lead at Megvii Technology Inc.
Haidian District, Beijing, China
Summary
👤
Senior🎓
Top SchoolXinran Xu is a Technical Lead with 13 years of experience building high-performance distributed systems, deep learning infrastructure, and full-stack services from Tsinghua University to production at Megvii. He has led teams of 10+ to design billion-scale face search and object storage systems, and currently drives MegEngine development for GPU training and embedded inference. A pragmatic polyglot (Go, Python, C++, JS), he specializes in rapid prototyping across frontend, backend and cluster-level IaaS/PaaS, including custom tooling for PXE/IPMI provisioning and a Redis-like small-file storage called Nori. His hands-on background spans low-level performance fixes—evident in contributions to the PyAV FFmpeg bindings—up to system-wide operational automation managing hundreds of GPU servers. Colleagues rely on him to turn research needs into production-grade platforms that prioritize availability, scale, and engineer operability.
13 years of coding experience
5 years of employment as a software developer
Bachelor’s Degree, Computer Science, Bachelor’s Degree, Computer Science at Tsinghua University
Stackoverflow
1reputation
0reached
0answers
0questions
Github Skills (4)
ffmpeg10
cython10
python10
clib9
Programming languages (10)
Github contributions (5)
PyAV-Org/PyAV
Jul 2015 - Feb 2016
Pythonic bindings for FFmpeg's libraries.
Role in this project:
Back-end Developer
Back-end DeveloperContributions:5 commits, 4 PRs in 7 months
Contributions summary:Xinran primarily contributed to the pyav project by addressing low-level details related to FFmpeg bindings. Their work included fixing return types, adding library directories for compilation, and resolving a refactoring error in the build setup. They also made an ad-hoc fix for an analyze limit and added a flag to improve the process of obtaining PTS (Presentation Time Stamps).
ffmpegpyavpythonpythonic
MegEngine/Hub
Nov 2020 - May 2022
基于旷视研究院领先的深度学习算法，提供满足多业务场景的预训练模型
Contributions:4 reviews, 5 commits, 3 PRs in 1 year 6 months
deep-learningmodelmachine-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.