Uber leverages data science to explore new frontiers and provide access to safe and reliable transportation for 75 million monthly active riders taking 15 million trips every day, providing opportunity for millions of drivers. Outside of ridesharing, we are also branching into new business areas, including Uber Eats, Uber Freight, and Uber Health, and modalities, including bikes and buses. Our massive scale and global presence bring unique and challenging problems, and with great problems comes the need for technical expertise, diverse perspectives, and an appetite for pushing the boundaries of what is possible with data science.
In this spirit of this charter, our Women in Statistics, Data, Optimization and Machine Learning (WiSDOM) employee resource group hosted an October 2018 meetup that celebrated the technical journeys and projects of some of the women at Uber who are moving the world with data through their work at Uber. During this inaugural meetup, members of WiSDOM, an organization for empowering women working in data science at Uber, shared their experiences at the company and discussed the opportunities they have to make meaningful product changes at scale by leveraging both traditional and cutting edge data science techniques.
Read on to learn more about how we do data science at Uber:
Food Discovery with Uber Eats: Recommending for the Marketplace
Uber Eats data scientist Yuyan Wang discusses how ranking and recommending restaurants on the Uber Eats platform is a unique challenge given the app’s three-sided marketplace of eaters, restaurant-partners, and delivery-partners. Yuyan walks through the technical journey of building the Uber Eats restaurant ranking system and goes into detail about how we developed a multi-objective optimization framework that takes into account the needs of the entire marketplace, as well as a holistic personalization model for heterogeneous and hierarchical content (in the case of Uber Eats, restaurant and dish recommendations).
An Intriguing Failing of Convolutional Neural Networks and the CoordConv Solution
Rosanne Liu walked us through her work with Uber AI Labs on the general inability of convolutional neural networks in transforming spatial representations between different types of coordinate systems. Her team’s innovative solution in the form of the CoordConv layer brought improvements in a wide range of domains, including object detection, generative models, and reinforcement learning.
Mediation Modeling at Uber
Bonnie Li discusses how she applies mediation modeling, a statistical approach from academic research, to address user pain points for the Uber Applied Behavioral Science team. Mediation modeling goes beyond simple cause and effect relationships in an attempt to understand what underlying mechanisms lead to a given result. Using Bonnie’s work, we can fine-tune product changes and develop new ones that focus on the underlying mechanisms behind successful features on the Uber platform.
To close out the event, Uber panelists, including data scientists Emily Bailey and Bonnie Li, along with data science directors Dawn Woodard and James Rauen, spoke to moderator Katherine Chen (Data Scientist), about topics close to their hearts: career advice for their younger selves, top data science book recommendations, and what problems they would focus on if they were Uber CEO Dara Khosrowshahi for a day.
Want to work on data science problems at Uber-scale with WiSDOM and other members of the Uber Data Science community? If any of the topics outlined above interests you, consider applying for a role on our team or joining our Meetup group to attend future Uber Engineering events.