Overview
Explore how Uber leverages Ray, a distributed compute platform, to address the challenges of large-scale deep learning training and tuning. Learn about the evolution of Uber's Michelangelo Machine Learning Platform from Apache Spark to Ray, driven by the mass adoption of deep learning techniques. Discover case studies highlighting specific infrastructure challenges in compute, network, and storage posed by deep learning, and understand how Ray's capabilities are utilized to overcome these obstacles. Gain insights into the transformation of Uber's technical architecture to support advanced machine learning applications that power critical business decisions and enhance customer experiences.
Syllabus
Large-scale deep learning training and tuning with Ray at Uber
Taught by
Anyscale