Marc Rasi
Software Engineer at Lutra AI
San Francisco Bay Area United States
Summary
🤩
Rockstar🎓
Top SchoolMarc Rasi is a software engineer with 12 years of experience in the San Francisco Bay Area, currently building at Lutra AI after multi-year engineering roles at Google and Coursera. He combines a strong mathematical foundation from Stanford with hands-on systems and ML work, contributing to high-profile open-source projects like Swift for TensorFlow and fast.ai where he implemented callback frameworks, optimizers, and a Swift Jupyter kernel. Comfortable across backend, full-stack, and ML library code, he has a track record of improving core tooling, build processes, and derivative/optimizer implementations that enable more flexible model training. His contributions show a knack for bridging low-level language/runtime work with practical developer-facing features, and for turning research-oriented code into robust, usable components.
12 years of coding experience
10 years of employment as a software developer
Bachelor of Science (BS), Mathematics, Bachelor of Science (BS), Mathematics at Stanford University
Stackoverflow
116reputation
4kreached
2answers
0questions
Github Skills (33)
loss-functions10
python10
optimizers10
model-driven10
machine-learning10
optimizer10
rnn-model10
n10
swift-package-manager10
lldb10
model-building10
deep-learning10
tensorflow10
swift10
jupyter-notebook10
Programming languages (10)
Github contributions (5)
tensorflow/swift-apis
Feb 2019 - Dec 2020
Swift for TensorFlow Deep Learning Library
Role in this project:
ML Engineer
ML EngineerContributions:1 release, 11 reviews, 28 commits in 1 year 10 months
Contributions summary:Marc primarily contributed to the Swift for TensorFlow deep learning library by implementing and refining core components related to optimizers and loss functions. Their work included adding initializers for optimizers like Adam, RMSProp, and SGD, along with integration of new loss functions, such as index-based softmax cross entropy. Additionally, they improved the library by including benchmarking tools and addressing derivative calculations, which involved fixing derivative implementations related to tensors and recurrent neural network layers.
deep-learningswift-for-tensorflowdifferentiable-programmingtensorflowswift
tensorflow/swift
Aug 2019 - Sep 2020
Swift for TensorFlow
Role in this project:Full-stack Developer
Full-stack DeveloperContributions:5 reviews, 27 commits, 56 PRs in 1 year
Contributions summary:Marc contributed to the development of a Swift Jupyter kernel, enabling Swift code execution within Jupyter notebooks. Their work included creating a Python-based kernel using the LLDB API for debugging and code evaluation. The user also created and updated installation scripts and notebooks for the Swift kernel, along with an example tutorial for using the kernel in Google Colab. Furthermore, they added license information and made various documentation and code updates.
swift-for-tensorflowmachine-learningdifferentiable-programmingtensorflow
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.