Sijun H
Shenzhen, Guangdong Province, China
Summary
🤩
Rockstar🎓
Top SchoolSijun H is an experienced ML engineer with 11 years focused on Deep Learning, NLP and content recommendation systems, currently based in Shenzhen and open to staff+ ML roles. He led NLP teams at Twitter, progressing from ML Engineer to Staff ML Engineer and Tech Lead, driving foundational language models and applied content-understanding features. At Baidu and on GitHub he contributes to large open-source NLP projects like PaddleNLP, notably improving Electra for efficient decoding and reliability in production settings. With a Stanford Master’s in Statistics and hands-on experience building LLMs and agents, he blends rigorous probabilistic thinking with practical engineering at scale. A less obvious strength is his track record of shipping core model architecture changes and verification (unit tests and caching correctness), which bridges research-grade models and production robustness.
11 years of coding experience
6 years of employment as a software developer
High School, High School at Westtown SchoolBachelor’s Degree, Mathematics, Bachelor’s Degree, Mathematics at Rose-Hulman Institute of TechnologyMaster’s Degree, Statistics, Master’s Degree, Statistics at Stanford University
Chinese, Chinese, English
Github Skills (6)
transformers10
paddlepaddle10
nlp10
python10
electra10
unit-testing9
Programming languages (8)
Github contributions (5)
PaddlePaddle/PaddleNLP
Oct 2022 - Jan 2023
Easy-to-use and powerful LLM and SLM library with awesome model zoo.
Role in this project:
ML Engineer
ML EngineerContributions:10 releases, 1250 reviews, 72 commits in 3 months
Contributions summary:Sijun primarily contributed to the implementation and support of the Electra model within the PaddleNLP framework, adding features such as the `past_key_values` argument for efficient decoding and `inputs_embeds` for enhanced flexibility. They also worked on adding unit tests and checking the model's caching mechanism to ensure correctness. These changes involved modifying the model's core architecture and training procedures.
extractionsemantic-similaritybertnlp-tasksfinetuning
sijunhe/ChineseCheckersAI
Apr 2016 - Jun 2016
Contributions:57 commits, 1 PR, 49 pushes in 2 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.