Abstract: 

In this talk we discuss the different demand forecasts we need to generate to optimize the interventions we make in the three-sided marketplace of on-demand food delivery. We go through different data preparation and modeling techniques for different prediction horizons and granularity. Further, we discuss some of the downstream secondary models which consume these forecasts to make optimal business decisions. Lastly we touch on how the Data Science team is organized at DoorDash to best execute on these problems.

Bio: 

As Head of Data Science and Machine Learning at DoorDash, Alok Gupta directs teams building models to optimize critical metrics necessary to balance a three-sided marketplace. With previous roles as Director of Data Science for both Lyft and Airbnb, Alok has abundant experience in using data science to make predictions at scale. Prior to this industry experience, Alok was a Research Fellow in Mathematics at Oxford University, earning his PhD in derivative pricing, work that he applied on Wall Street as a high frequency trader to predict movements in foreign exchange rates at high frequency.