Weikeng Q
Member Of Technical Staff at OpenAI
Mountain View, California, United States
Summary
🤩
Rockstar🎓
Top SchoolWeikeng Q is a Member of Technical Staff and seasoned software engineer with a decade of experience building high-throughput, distributed services and production ML systems from Mountain View. He combines deep systems and infrastructure expertise (Java/C++/Python, MySQL and NoSQL, memcache, Tomcat/Spring) with research-grade contributions to speech and representation learning—leading work on Conformer, large-scale SSL, and multimodal AudioPaLM/Gemini during a long tenure at Google DeepMind. He’s a core contributor to production ML frameworks like Lingvo and Praxis and has hands-on experience fixing low-level TensorFlow/CuDNN compatibility and inference issues in widely used repos. At Google and Amazon he designed scalable, real-time services deployed to high-QPS products (Meet, YouTube captions, Pixel) and now applies that hybrid research-to-production skillset at OpenAI. Notably, his background spans both algorithmic foundations and pragmatic engineering—optimizing performance, reproducibility, and scalability across models trained on millions of hours of audio.
10 years of coding experience
10 years of employment as a software developer
Bachelor of Engineering (BEng), Information Engineering, 91.6/100, Bachelor of Engineering (BEng), Information Engineering, 91.6/100 at Southeast UniversityMaster of Engineering, Computer Engineering, 3.9/4.0, Master of Engineering, Computer Engineering, 3.9/4.0 at University of Virginia
Github Skills (12)
deep-learning10
tensorflow10
conformer10
python10
speech-recognition9
machine-learning9
nlp9
architecture8
architectures8
cuda7
cprogramming-language5
c-language5
Programming languages (3)
Github contributions (5)
tensorflow/lingvo
Nov 2018 - Sep 2022
Lingvo
Role in this project:
Back-end Developer & ML Engineer
Back-end Developer & ML EngineerContributions:138 commits, 2 comments in 3 years 11 months
Contributions summary:Weikeng primarily focused on updating and maintaining the Lingvo codebase, specifically addressing compatibility issues with TensorFlow's CuDNN interface and making minor improvements to the predictor module. Furthermore, they improved shape inference within the SpeechOnlineEncoderV1 module and corrected a potential freezing issue in the inference graph exporter. The user also appears to work on refactoring and adding new features to the Conformer architecture implementation.
asrtranslationctcspeech-recognitiontensorflow
protoget/dotfiles
Jan 2019 - Sep 2019
Contributions:12 pushes, 1 branch in 7 months
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.