Austin Jacobson
AI Research Scientist at MosaicML
San Francisco, California, United States
Summary
🤩
Rockstar🎓
Top SchoolAustin Jacobson is an AI research scientist with 11 years of experience building and deploying machine learning and deep learning systems across images, text, speech, and structured data. Based in San Francisco, he blends research-grade experimentation with production engineering from roles at MosaicML and Databricks, and prior hands-on ML work at Zesty.ai and as a consultant. He contributes to open-source ML tooling—helping extend MosaicML’s Composer with label-smoothing fixes and timm model support—demonstrating a knack for integrating pre-trained ecosystems into training frameworks. Trained originally in finance, he brings a pragmatic, data-driven perspective to model evaluation and product impact, often bridging research prototypes to scalable workflows.
11 years of coding experience
5 years of employment as a software developer
Bachelor’s Degree, Finance, Bachelor’s Degree, Finance at University of North Florida
Github Skills (9)
neural-network10
pytorch10
machine-learning10
deep-learning10
integrate10
integrations10
data-integration10
ml10
python9
Programming languages (7)
Github contributions (5)
mosaicml/composer
Nov 2021 - Oct 2022
Supercharge Your Model Training
Role in this project:
ML Engineer
ML EngineerContributions:142 reviews, 26 commits, 49 PRs in 10 months
Contributions summary:Austin contributed to the `mosaicml/composer` repository, which focuses on supercharging model training, as indicated by the repository description. The user made changes related to label smoothing, including fixing an issue where it would not work without alpha set and modifying an example file to include alpha. The user also added support for the timm library, which provides pre-trained models, by creating new files and modifying existing ones to support timm models, further demonstrating an effort to integrate the model with the framework.
pytorchml-systemsdeep-learningneural-networksmachine-learning
A-Jacobson/shiba
May 2019 - Mar 2022
A simple, flexible, pytorch training loop.
Contributions:95 commits, 12 PRs, 84 pushes in 2 years 10 months
nlppytorchdeep-learningcomputer-visionloop
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.