By Wayne Cunningham
Uber held its first open source summit on November 15, 2018, inviting members of the open source community for presentations given by experts on some of the projects we have contributed in the fields of big data, visualization, machine learning, and front-end web application building.
Kicking off Uber Open Summit 2018, Jim Zemlin, Executive Director of the Linux Foundation, delivered a keynote outlining the benefits of open source software and the many industries that make use of it. Uber AI Chief Scientist Zoubin Ghahramani closed out the day with a final keynote covering the advances and challenges of artificial intelligence.
The one-day event, held in San Francisco, consisted of three tracks filled with instructive presentations, including practical tips on how to apply open source projects such as Fusion.js, Horovod, Pyro, Kepler.gl, H3, and Jaeger. Presenters from Uber, such as Senior Research Engineer Fritz Obermeyer, Software Engineer Nadiia Dmytrenko, Head of Visualization Nicolas Garcia Belmonte, and Senior Software Engineer Xiaoji Chen, shared their knowledge built from working with these projects in a production environment on a daily basis.
The following selection of videos gives an overview of Uber Open Summit 2018. Please check out the Uber Engineering Youtube channel for more from our event.
Keynote with Jim Zemlin
Highlighting the news of Uber’s Gold membership in the Linux Foundation, Executive Director Jim Zemlin led the Uber Open Summit with the first keynote of the day, showing how open source software benefits and enables an increasingly wide array of industries, along with the role of newly-minted open source program offices at many companies.
Pyro – Deep Universal Probabilistic Programming
Senior Research Engineer Fritz Obermeyer discusses how Pyro works as a platform for modern, Bayesian machine learning, and its rich, stable toolset for probabilistic machine learning research developed by the open source community.
Scaling a React Component Library – Building Uber’s New Design Language
Software Engineer Nadiia Dmytrenko gives a foundational talk about how Uber uses React in our UI platform, defining design patterns and the anatomy of components.
Urban Computing with Advanced Visualization
Head of Visualization Nicolas Garcia Belmonte relates the ways in which open source visualization technologies developed at Uber can be used to understand urban phenomena or even predict the future of cities.
Large-Scale Geospatial Visualization in the Browser with deck.gl
Senior Software Engineer Xiaoji Chen gives a hands-on workshop about deck.gl’s core concepts and capabilities via real-world examples, showing how this WebGL-powered data visualization framework can help us understand transportation in our cities.
Horovod: Distributed Deep Learning in 5 Lines of Python
Senior Software Engineer Travis Addair shows how to take a single-GPU training script written in any deep learning framework, and scale it to efficiently run on hundreds of GPUs with just a few lines of code on Horovod.
Petastorm: Training and Evaluation of Deep Learning Models Directly from Datasets in Apache Parquet Format
Senior Software Engineer Yevgeni Litvin of Uber ATG explains the difficulty of working with large datasets derived from sensors, and how Petastorm makes this data available to deep learning models.
Keynote with Zoubin Ghahramani
Delivering the closing keynote, Uber AI Chief Scientist Zoubin Ghahramani relied on his vast experience with artificial intelligence to not only show what it can do, but a few areas in which it struggles. Uber contributes to machine learning through research and open source projects, including Pyro and Horovod.
Want to get involved? If any of the topics outlined above interests you, consider applying for a role on our team, and if you would like to attend future Uber Engineering events, join our Meetup group.
Subscribe to our newsletter to keep up with the latest innovations from Uber Engineering.