Jason Mancuso
Member Of Technical Staff, Research at Modal
New York, New York, United States
Summary
🤩
Rockstar🎓
Top SchoolJason Mancuso is a Forward Deployed ML Engineer with nine years of experience applying research-driven machine learning to real-world problems, currently based in New York. He has deep expertise in federated and privacy-preserving ML, demonstrated by substantive contributions to high-profile open-source projects like TensorFlow Federated and OpenMined/PySyft where he redesigned executor strategies and integrated remote tensor/ autograd workflows. At Cape Privacy he translated privacy-aware research into production-ready features, and he now brings that blend of research and deployment focus to Modal. Jason combines back-end engineering chops with a strong mathematical foundation (BS in Mathematics) and a track record founding and organizing community efforts like Cleveland AI. Colleagues rely on him to bridge complex ML theory and scalable systems, especially when models must operate on decentralized or sensitive data. He often surfaces non-obvious system-level improvements—rewriting core execution logic or autograd paths—that materially increase capability and robustness.
9 years of coding experience
10 years of employment as a software developer
Bachelor of Science (BS) Mathematics, Bachelor of Science (BS) Mathematics at John Carroll University
Stackoverflow
1reputation
0reached
0answers
0questions
Github Skills (18)
asynchronous10
pytorch10
operation10
python10
async10
tensorrt10
machine-learning10
asyncio10
deep-learning10
tensorflow10
federated-learning10
autograd10
tensor10
sym10
testing9
Programming languages (12)
Github contributions (5)
OpenMined/PySyft
Apr 2018 - Oct 2019
Perform data science on data that remains in someone else's server
Role in this project:
Back-end Developer & ML Engineer
Back-end Developer & ML EngineerContributions:72 commits, 44 PRs, 37 pushes in 1 year 6 months
Contributions summary:Jason implemented variable integration, including sending and receiving tensor objects, within the `pysyft` framework. They modified services to handle and transport variable data, including remote matrix multiplications. Furthermore, the user was involved in fixing autograd functionality and creating a demo for the Denver ML Grid, indicating a focus on integrating and debugging deep learning capabilities, specifically for federated and privacy-preserving machine learning. They worked on changes related to autograd to make it work with larger computation graphs.
data-sciencedeep-learningsecure-computationpytorchprivacy
google-parfait/tensorflow-federated
Mar 2020 - Jul 2020
An open-source framework for machine learning and other computations on decentralized data.
Role in this project:Back-end Developer / ML Engineer
Back-end Developer / ML EngineerContributions:12 commits, 7 PRs, 29 comments in 3 months
Contributions summary:Jason primarily focused on enhancing the `tensorflow-federated` repository by implementing and refactoring core components within the executor framework. They introduced an `IntrinsicStrategy` and moved intrinsic logic to a `CentralizedIntrinsicStrategy`. Their work included modifications to executors, introducing and testing new intrinsic strategies, and addressing issues related to unsupported intrinsics. These changes suggest a deep understanding of the project's architecture and its capabilities for machine learning on decentralized data.
machine-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.