Jianyu Wang
Research Scientist at Apple
Cupertino, California, United States
Summary
👤
Senior🎓
Top SchoolJianyu Wang is a research scientist with 11 years of experience building frontier foundation models and optimizing LLM pretraining, architectures, and on-device inference from Cupertino. With a PhD student background at Carnegie Mellon and prior roles across Apple, Meta, and Google, he combines deep academic rigor with hands-on production research in federated and privacy-preserving ML. His contributions to Google Research’s federated learning codebase include optimizer innovations, dataset pipeline improvements for CIFAR-10, and training loop refactors that improved efficiency on resource-constrained clients. Jianyu’s work emphasizes adaptive optimizers and communication-efficient distributed training—skills he applied during internships at FAIR and Google before transitioning to industry research. Known for bridging algorithmic advances with practical engineering, he brings uncommon expertise in making large models performant on-device and in federated settings.
11 years of coding experience
1 year of employment as a software developer
Bachelor of Engineering - BE, Electronic Engineering, Bachelor of Engineering - BE, Electronic Engineering at Tsinghua UniversityDoctor of Philosophy - PhD, Electrical & Computer Engineering, Doctor of Philosophy - PhD, Electrical & Computer Engineering at Carnegie Mellon University
Github Skills (10)
tensorflow10
python10
federated-learning10
optimization9
optimizers9
optimisation9
machine-learning9
algorithm9
data-preprocessing8
keras7
Programming languages (1)
Github contributions (5)
google-research/federated
Dec 2020 - Mar 2021
A collection of Google research projects related to Federated Learning and Federated Analytics.
Role in this project:
ML Engineer
ML EngineerContributions:3 reviews, 5 commits, 2 PRs in 2 months
Contributions summary:Jianyu implemented and refined federated learning algorithms within the Google Federated project. Their contributions involved modifying the dataset loading process for CIFAR-10, introducing a joint correction method for local adaptive optimizers, and refactoring code to align with Google's style guide. These changes reflect a focus on improving the efficiency and performance of federated learning models. The user also made changes to the training loop configuration.
federated-learning
JYWa/Overlap_Local_SGD
Feb 2020 - Jul 2020
Implementation of (overlap) local SGD in Pytorch
Contributions:18 commits, 17 pushes, 1 branch in 4 months
pytorchdistributed-optimizationsgd-optimizerdeep-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.