Linxiao Zeng is a Machine Learning Engineer based in Singapore with 8 years of experience specializing in NLP and machine translation, particularly for Asian (CJK) languages. Currently building real-time speech translation services at Zoom, he brings production-focused expertise in translation pipelines, tokenization, and constraint handling. As a key contributor to the widely used OpenNMT-py project, he enhanced REST server capabilities, n-best support, and subword vocabulary handling—work that bridges research advances and deployable systems. His background spans research and product roles where he applied BERT for quality boosts, cut post-editing by 70% through novel constraint techniques, and explored document-level translation and end-to-end speaker diarization. Educated in engineering and machine learning in China and France, he combines strong academic foundations with hands-on system building across research and production contexts.
8 years of coding experience
3 years of employment as a software developer
Bachelor of Engineering - BE, Electrical and Electronics Engineering, 3.88/4.00, Bachelor of Engineering - BE, Electrical and Electronics Engineering, 3.88/4.00 at Huazhong University of Science and Technology
Master of Engineering - MEng, Computer Science, Master of Engineering - MEng, Computer Science at CentraleSupelec
Master of Science - MS, Machine Learning, Master of Science - MS, Machine Learning at Université Paris-Saclay
Open Source Neural Machine Translation and (Large) Language Models in PyTorch
Role in this project:
Back-end Developer
Contributions:32 reviews, 39 commits, 39 PRs in 2 years 2 months
Contributions summary:Linxiao primarily contributed to the REST server functionality within the OpenNMT-py repository. Their work included adding and modifying preprocessor and postprocessor options, enhancing the server's support for tokenization, and enabling support for n-best translations. The contributions focused on improving the server's capabilities for translation and processing, with additions such as target prefix support, and improvements related to the handling of subword vocabularies.
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.