Xiaohui Zhang
Staff Research Scientist at Meta
New York, New York, United States
Summary
🤩
Rockstar🎓
Top SchoolXiaohui Zhang is a Staff Research Scientist based in New York with 13 years of experience building production speech and audio systems, currently leading Monetization GenAI efforts at Meta. She has driven technical roadmaps and led teams to deliver voice generation and audio capabilities for MetaAI and AI Characters, while remaining hands-on in model and pipeline development. Her background includes founding and tech-leading Facebook’s Video ASR team that deployed multilingual ASR across Facebook and Instagram, and steering TorchAudio’s direction to benefit both Meta and the open-source community. A Johns Hopkins Ph.D. with deep contributions to the widely used Kaldi ASR toolkit, she combines rigorous research in acoustic modeling, optimization, and model compression with practical script-level engineering for signal processing pipelines. Equally comfortable in manager and individual-contributor roles, she blends leadership, research pedigree, and an uncommon knack for evolving low-level audio tooling into scalable, production-grade systems.
13 years of coding experience
7 years of employment as a software developer
Johns Hopkins UniversityBeijing No.2 Middle SchoolBachelor of Engineering (BEng), Electronic Engineering, minor in Applied Physics, 92.2/100, Bachelor of Engineering (BEng), Electronic Engineering, minor in Applied Physics, 92.2/100 at Beijing University of Technology
Github Skills (10)
sh10
kaldi10
script10
speech-recognition10
shell10
scripting10
speech-to-text9
c-language7
cprogramming-language7
cuda5
Programming languages (4)
Github contributions (5)
kaldi-asr/kaldi
Sep 2015 - Aug 2020
kaldi-asr/kaldi is the official location of the Kaldi project.
Role in this project:
Back-end Developer
Back-end DeveloperContributions:56 PRs, 280 comments in 4 years 11 months
Contributions summary:Xiaohui's commits primarily involve modifying and adding features to shell scripts within the Kaldi project. These changes relate to audio processing steps, particularly those involving Gaussian selection, lattice generation, and MMI/sMBR training. The contributions include the addition of new functionality, fixing logical errors within existing scripts, and general code maintenance. A focus on modifying shell scripts used for signal processing and acoustic modeling, is apparent in the user's commits.
cudakaldiasrspeech-to-textkaldi-asr
xiaohui-zhang/kaldi
Sep 2015 - May 2019
This is now the official location of the Kaldi project.
Contributions:199 pushes, 24 branches in 3 years 8 months
kaldilocation
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.