Driving Uber: ORIE makes its mark in ridesharing technology
By Olivia M. Hall
Cornell grads are a driving force at transportation technology company Uber. Over the past couple of years, nearly a dozen alumni (mostly from ORIE) as well as former and current ORIE faculty members have made their way to the corporation’s San Francisco headquarters to help shape the future of ridesharing.
“It is an opportunity to work with top-notch talent on the most challenging analytic problems in the world,” said Bob Phillips, until recently director of the Marketplace Optimization Data Science (MODS) group, where most of the Cornelians are clustered.
As ridesharing has caught on and the company has expanded its offerings to include food and package delivery in addition to carpooling and other projects, the company has grown rapidly. MODS now boasts 100 data scientists, of whom the Cornell members make up a disproportionate 10 percent (considering that Uber’s natural recruiting grounds are West Coast universities).
ORIE alumni have made their mark throughout the group’s teams, frequently in leadership positions, contributing to “the data science and analytics behind all the core problems powering Uber’s business,” Phillips said.
Take UberPOOL, a carpooling service in which riders trade a cheaper fare for sharing the vehicle with others and taking a little longer to reach their destination. Lior Seeman Ph.D. ’15 helms the effort to keep that time to a minimum by “improving the pool matching algorithm to try and get as many people as possible in one car travelling to the same area,” the computer scientist explained. Steve Pallone Ph.D. ’17 has recently also joined the Matching team.
A new feature in the Uber app gives customers an estimate when they will arrive in order to reduce their uncertainty, though the team learned that this is only one of several factors riders consider: “Interestingly, they care a lot about the number of stops, number of people in the vehicle, and whether it travels backwards,” said ORIE associate professor and Uber staff scientist Peter Frazier, who led the UberPOOL data science team for a year and a half and now splits his time between Ithaca and the West Coast.
Mike Freimer Ph.D. ’01 heads up the control team, which tries to understand the health of cities as Uber marketplaces, as well as the Forecasting team, “providing forward-looking projections of supply (available drivers) and demand (riders) to our pricing engines,” he said. “These forecasts are generated across multiple levels of spatial and temporal granularity.”
Finally, Dawn Woodard, a former associate professor in ORIE and the Department of Statistics, has moved to the top of data science for Uber Maps after previously leading the Dynamic Pricing team, of which James Davis Ph.D. ’15 is a member. “We set prices in an effort to keep the marketplace healthy,” Davis explained. Riders know up front how much they will pay for a trip, with fares adjusting up or down according to demand.
This pricing model is fundamental to Uber’s business, Frazier contends. “You can argue that Dawn’s work was a lynchpin for Uber’s success, and I believe we [Cornelians] have played a critical role in bringing the company to where it is today.”
Conversely, the ability to have such a tangible, real-world effect makes Uber an attractive employer. “It’s a place that’s unique in having the speed of a startup company but the scale of a global company,” Frazier said.
So when Davis designs a small algorithmic change, it can positively influence the earnings of drivers world-wide. “I sought out a job with interesting and impactful problems,” he said. “Uber fit the bill and now I feel like I have a strong voice in designing the future of ridesharing.”
Because it is a relatively new product, ridesharing continues to evolve and present complex problems with a physical component—moving people through cities in cars—that sets Uber apart from other technology companies. It also ties into OR’s historical emphasis on addressing practical issues in military, manufacturing, or transportation applications.
ORIE graduates thus feel well prepared to tackle the dynamic challenges of ridesharing. “The OR department provided me with an excellent technical toolbox,” Freimer said. “Optimization, statistics, stochastic modeling and simulation have all been directly relevant. Also hugely important was practice tackling ambiguous, ill-defined problems and creating analytical frameworks for addressing them.”
Phillips agrees: “Cornell Ph.D. grads tend to have a highly pragmatic approach,” he said. “They are interested in solving problems rather than proving theorems. They are not afraid to get their hands dirty and they seem to love working with data. This is a huge plus at Uber.”
And thus word of mouth continues to bring Cornell grads to San Francisco. “Uber is simply the most exciting OR company around,” Frazier said.
Headquarters: San Francisco, Calif.
Daily trips: 10 million
Riders: 40 million
Drivers: 2 million