Jesse Cai
Machine Learning Engineer at Meta
San Francisco, California, United States
Summary
🤩
Rockstar🎓
Top SchoolJesse Cai is a Machine Learning Engineer with 11 years of experience building production ML systems and optimizing model architectures, currently focused on sparsity and quantization for PyTorch at Meta. He has a strong track record shipping NLP and representation-learning solutions—leading multilingual, privacy-minded pipelines and seq2seq feedback assistants at Cultivate and deploying LSTM/RNN models and Spark ingestion at Blend. Jesse is an active open-source contributor to the flagship pytorch/pytorch repository, where his backend work on quantization, fp8 support and sparse tensor compilation has simplified internals and improved performance. He combines research instincts from UCLA with hands-on product delivery in San Francisco, and brings a practical flair for bridging cutting-edge model techniques with scalable engineering.
11 years of coding experience
3 years of employment as a software developer
Bachelor of Science - BS, Computer Science, Bachelor of Science - BS, Computer Science at University of California, Los Angeles
Stackoverflow
41reputation
7kreached
2answers
1question
Github Skills (21)
pytorch10
python10
operation10
tensorrt10
sparse-matrix10
machine-learning10
tensorflow10
cuda10
quantization10
tensor10
sparse-data10
back-end-development9
decimal6
unsigned-integers6
operating-system6
Programming languages (9)
Github contributions (5)
pytorch/pytorch
Aug 2022 - Jan 2023
Tensors and Dynamic neural networks in Python with strong GPU acceleration
Role in this project:
Back-end Developer & ML Engineer
Back-end Developer & ML EngineerContributions:1 release, 167 reviews, 233 commits in 5 months
Contributions summary:Jesse's commits primarily focus on enhancing the PyTorch framework's capabilities, particularly within the domain of quantization. They've made substantial changes by removing index dictionaries and implementing input type mappings to the backend patterns for functions like `nn.functional.linear` and `nn.functional.conv1d`, which simplifies the codebase. The user has also added and refined support for various operations and data types, including extending support to fp8 datatypes, as well as made improvements to the cuSPARSELt backend, which is a core part of the sparse tensor support. They are also responsible for adding torch compile support to the sparse semi structured tensor implementation.
gpu-accelerationneural-networkpythonautogradgpu
jcaip/configs
Oct 2015 - Jun 2022
Contributions:6 pushes, 1 branch in 6 years 8 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.