Xingjian Shi
Member Of Technical Staff at OpenAI
California, United States
Summary
🤩
Rockstar🎓
Top SchoolXingjian Shi is a deep learning researcher and engineering leader with 11 years of experience building multimodal foundation models, currently advancing multimodal research at OpenAI. He co-founded and led model development at Boson AI, shipping the Higgs series (Higgs Llama and Higgs Audio) that span text and audio modalities. Previously at AWS he led AutoGluon Multimodal—extending AutoML with foundation models—and contributed to DeepEarth, an ambitious effort to build Earth-scale foundation models. His open-source footprint includes substantive contributions to high-profile ML projects like autogluon, mxnet, tvm and gluon-nlp, where he improved NLP workflows, core operators and cross-platform compatibility. With a PhD from HKUST and a background in low-level ML libraries (mshadow) and compiler/operator work, he bridges research, production engineering, and system-level optimization. Based in California, he combines academic rigor with startup-driven product delivery and a knack for making foundational tooling more practical and performant.
11 years of coding experience
6 years of employment as a software developer
Shanghai Jiao Tong UniversityHong Kong University of Science and Technology (HKUST)
Github Skills (44)
beam-search10
c-language10
matrix10
operation10
python10
tensorrt10
mxnet10
gpu-programming10
machine-learning10
test-framework10
gluon10
numpy10
mat10
compiler-compiler10
deeplearning-ai10
Programming languages (8)
Github contributions (5)
dmlc/gluon-nlp
Apr 2018 - Feb 2021
NLP made easy
Role in this project:
ML Engineer
ML EngineerContributions:223 reviews, 62 commits, 180 PRs in 2 years 10 months
Contributions summary:Xingjian primarily contributed to the development of the `gluon-nlp` repository, focusing on improving and adding features related to natural language processing. Their work included fixing issues in the bucket sampler, adding beam search capabilities, and integrating attention mechanisms. The user also implemented and tested the integration of a machine translation dataset.
nlpmxnetnatural-language-understandingnlgnatural-language-generation
autogluon/autogluon
Feb 2020 - Jan 2023
Fast and Accurate ML in 3 Lines of Code
Role in this project:Back-end Developer & ML Engineer
Back-end Developer & ML EngineerContributions:1044 reviews, 81 commits, 272 PRs in 2 years 11 months
Contributions summary:Xingjian's commits primarily involved adding dependencies and refactoring code related to Natural Language Processing (NLP) tasks within the AutoGluon framework. They introduced a dependency on the pynvml library, likely for monitoring GPU usage, and updated the setup.py file. The refactoring efforts focused on revising the dependency structure of gluonnlp to lazy dependency, which suggests optimization for loading and utilizing NLP modules. These contributions are centered around enhancing the Autogluon's capabilities in machine learning and text processing, particularly around NLP workflows and frameworks.
forecastingimage-textmlppythonmeta-learning
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.