Building a Uber experience

Uber has developed many great products but for every rider at any given time there is only one experience, and we need to make it seamless and truly differentiating. At Uber Marketplace, we are a team called Personalization and our mission is to provide the best product experience to the right rider at the right time. Built on top of Uber’s amazing big data infrastructure, our deep learning solution “learns” and “adapts” to rider preferences, working tirelessly in the background to minimize friction points, and pulls in various solutions to delight and excite our riders. The goal is to provide best possible experience to our riders. The core of this work is our rider targeting model, which is a machine learning model built on top of a number of rider features. We need to be able to understand our riders and their preference. The data pipeline is fully automated, the model is automatically trained, validated and deployed. We keep exploring new features, new models, new experiences and features. What we provide: Data is the most important thing in any machine learning application. Uber has huge amount of data from huge number of riders from hundreds of cities in whole world. A solid and flexible platform makes trying new ideas much easier. For example, creating a new model, plugging the model into production, creating a new test campaign and measure it with A/B testing are all made easy. There is a group of data scientist and engineers with strong machine learning background working on this fun problem. We have fast iterations on new ideas. You can get an idea, implement it and test it in production in a couple weeks.