Speaker "Mingshi Wang" Details Back



Complex Machine Learning Workflows at Scale


Uber's Michelangelo is a machine learning platform that supports training and serving thousands of models in production. In this talk, we will describe how Michelangelo evolved its architecture to handle more complex machine learning workflows e.g. model ensembling.


My name is Mingshi Wang, a senior software engineer at Uber. I am currently working on Michelangelo - an internal machine learning platform at Uber - to help teams across the whole company to solve their business problems using cutting edge machine learning technologies. I am passionate in developing systems that reliably distribute the most commonly used open source machine learning frameworks at Uber's scale. My experience with Michelangelo has been exciting as I see Michelangelo grows from an immature system to a full fledged infrastructure of many of the most important projects at Uber. These projects includes driver pickup time ETA, Uber EAT restaurant ranking and the autopilot system for self-driving cars, to name a few.