Presentation: Michelangelo - Machine Learning @Uber

Michelangelo is the Machine Learning Platform that powers most of the machine learning solutions at Uber. The early goal of Michelangelo was to enable teams around Uber to deploy and operate ML solutions at Uber scale. Now that the system has matured, our focus has shifted towards developer velocity and empowering the individual model owners to be fully self-sufficient from early prototyping through full production deployment and operationalization. For data science users, we are making the operational side self-service so that they don’t need to engage engineering teams to deploy and operate the models and feature pipelines. For engineering users, we are making the modeling work easier and more automatic so that engineers can do more alone, without formal data science assistance.