Allen Shi
Machine Learning Research Engineer at Apple
Pittsburgh, Pennsylvania, United States
Summary
👤
Senior🎓
Top SchoolAllen Shi is a Machine Learning Research Engineer with 10 years of experience applying deep learning to large-scale personalization and ads systems across Amazon, Prime Video, and now Apple Intelligence. He holds a B.S. from Peking University and an M.S. from Carnegie Mellon and combines research rigor with production engineering—shipping models for retrieval, relevance, response prediction and homepage personalization. His open-source contributions include improving Transformer and BERT components in the texar-pytorch project, reflecting hands-on expertise in NLP model internals and PyTorch device handling. At Amazon he worked on core Sponsored Brand Advertising science and at Prime Video he leveraged heterogeneous behavior data to drive personalization, demonstrating an ability to translate academic ideas into measurable product impact. Known for bridging theory and systems, he has a track record of improving model robustness and integrating complex ML pipelines end-to-end.
9 years of coding experience
7 years of employment as a software developer
Bachelor's Computer Science, Bachelor's Computer Science at Peking UniversityMaster of Computational Data Science Computer Science, Master of Computational Data Science Computer Science at Carnegie Mellon University
Chinese, English
Github Skills (8)
pytorch10
machine-learning10
transformer10
gpt10
natural-language-processing9
bert9
python9
deep-learning9
Programming languages (6)
Github contributions (5)
asyml/texar-pytorch
May 2019 - Jul 2019
Integrating the Best of TF into PyTorch, for Machine Learning, Natural Language Processing, and Text Generation. This is part of the CASL project: http://casl-project.ai/
Role in this project:
ML Engineer
ML EngineerContributions:24 commits, 8 PRs, 2 branches in 1 month
Contributions summary:Allen made several commits focused on fixing model configurations, module imports, and initialization issues, particularly related to GPT-2 and the Transformer model. They added basic logic for the Transformer model, including decoding routines, and addressed device-related issues with PyTorch tensors. The contributions also involved adding and refining Transformer model components, along with fixing issues in BERT-related examples.
dialog-systemspythonxlnetbertlanguage-processing
haoransh/texar
Oct 2018 - Feb 2019
Toolkit for Text Generation and Beyond
Contributions:28 pushes, 3 branches in 4 months
nlpnatural-language-processingbeyondtext-generation
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.