Qiantong Xu
Staff Research Scientiest at Google DeepMind
Menlo Park, California, United States
Summary
🤩
Rockstar🎓
Top SchoolQiantong Xu is a Staff Research Scientist based in Menlo Park with a decade of experience building and optimizing ML systems across industry-leading AI labs and startups. He has driven research-to-production work at DeepMind, Character.AI, SambaNova, and Facebook AI, blending low-level C++ toolkit development (notably substantial contributions to the flashlight ML library) with system-level engineering on reconfigurable hardware and multimodal LLM audio pipelines. His background spans speech and language modeling, decoder performance optimization, and scalable dataflow architectures, reflecting both research depth and production rigor. A Cornell M.S. and Peking University alumnus, he often operates at the intersection of research and engineering—fixing memory explosions in decoders one moment and leading hardware-aware model deployments the next.
10 years of coding experience
7 years of employment as a software developer
Bachelor of Science (B.S.) Machine Intelligence, Bachelor of Science (B.S.) Machine Intelligence at Peking UniversityMaster's degree Computer Science, Master's degree Computer Science at Cornell University
Chinese, English
Github Skills (11)
modular-design10
machine-learning10
c-language10
deep-learning10
modular10
cprogramming-language10
performance-optimization9
algorithm9
algorithms9
autograd8
language-modeling8
Programming languages (3)
Github contributions (5)
flashlight/flashlight
Dec 2018 - Jun 2021
A C++ standalone library for machine learning
Role in this project:
Back-end Developer
Back-end DeveloperContributions:1 review, 85 commits, 31 PRs in 2 years 5 months
Contributions summary:Qiantong primarily contributed to the implementation and introduction of various modules and functionalities within the "flashlight/flashlight" repository, a C++ machine learning library. They introduced `UnaryModule`, `BinaryModule` and their corresponding `Module.h` and `modules.rst` documentation. Additionally, the user worked on implementing data preparation scripts for WSJ language modeling and included code changes related to decoder performance, specifically memory explosion issues. The commits demonstrate a focus on extending the library's capabilities and optimizing its performance.
cppdeep-learningstandalone-libraryc-plus-plusmachine-learning
xuqiantong/GAN-Metrics
Jun 2018 - Jun 2019
An empirical study on evaluation metrics of generative adversarial networks.
Contributions:20 commits, 3 PRs, 17 pushes in 11 months
pytorchinformation-theoryevaluation-metricsdeep-learningadversarial
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.