Yang Liu
Member Of Technical Staff at Microsoft AI
Redmond, Washington, United States
Summary
🤩
Rockstar🎓
Top SchoolYang Liu is an NLP researcher and Member of Technical Staff at Microsoft in Redmond, with eight years of experience building and fine-tuning large language models and production SWE agents. He holds a PhD from the University of Edinburgh and has progressed from research intern roles at AI2 and MSRA to senior and principal researcher positions focused on customized and post-training of GPT-class models. Yang is the author of widely used summarization toolkits like BertSum and PreSumm and led development on GPT-4-Japanese, combining deep academic grounding with pragmatic engineering. He balances research rigour with production delivery—optimizing training pipelines, model architectures, and deployment workflows—to move frontier LLM innovations into scalable systems.
8 years of coding experience
5 years of employment as a software developer
Doctor of Philosophy - PhD Natural Language Processing, Doctor of Philosophy - PhD Natural Language Processing at The University of EdinburghMaster of Science - MS Natural Language Processing, Master of Science - MS Natural Language Processing at Peking UniversityBachelor of Engineering - BE Computer Science, Bachelor of Engineering - BE Computer Science at Tianjin University
Github Skills (12)
text-summarization10
model-building10
pytorch10
machine-learning10
nlp10
deep-learning10
trainings10
python10
natural-language-processing10
modeling10
bert10
tensorboard7
Programming languages (4)
Github contributions (5)
nlpyang/BertSum
Mar 2019 - Aug 2019
Code for paper Fine-tune BERT for Extractive Summarization
Role in this project:
ML Engineer
ML EngineerContributions:75 commits, 34 pushes, 1 branch in 5 months
Contributions summary:Yang primarily contributed to the development and maintenance of the `bertsum` repository, which focuses on fine-tuning BERT for extractive summarization. Their commits involve modifying training procedures, data preprocessing steps, and model architecture, indicating a focus on refining the model's performance and adapting it to specific datasets. They also worked on integrating and configuring the BERT model, including loading pretrained weights and adjusting various configurations.
nlpdistilbertbertrobertafine
nlpyang/PreSumm
Aug 2019 - May 2020
code for EMNLP 2019 paper Text Summarization with Pretrained Encoders
Role in this project:ML Engineer
ML EngineerContributions:42 commits, 2 PRs, 67 pushes in 9 months
Contributions summary:Yang contributed to the core code of a text summarization project, primarily modifying model building and training scripts. They implemented modifications to the BERT model, specifically related to handling maximum positional embeddings. The user also made changes to the training and testing procedures, adjusting parameters like batch sizes. They introduced modifications for extractive summarization models, indicating involvement in both abstractive and extractive summarization methods.
nlprobertatext-summarizationencoderssummarization
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.