Speaker: Simon Ejdemyr, Netflix
Location: ISI Marina del Rey, Conference Room 1135. In-person attendance for USC-ISI faculty, staff, and students only. Open to the public virtually via Zoom.
Zoom meeting ID: 934 0714 5597
Hosted by: Myrl Marmarelis
POC: Karen Lake
Abstract: Netflix aims to pioneer the application of data science and artificial intelligence to enhance both the member experience and business outcomes. This presentation will focus on causal inference at Netflix, an area critical to understanding the impact of new features, algorithms, and content on our diverse user base. I will provide an overview of the landscape of (quasi-) experimentation in an online industrial environment, highlighting its relevance to both machine learning and generative AI. The discussion will then shift to two significant case studies that showcase the challenges and innovations in this setting: proxy metrics as a means to estimate otherwise insensitive business outcomes, and evaluation of the cumulative effects of a testing program after accounting for the winner’s curse.
This event will not be recorded.
Bio: Simon Ejdemyr is a Staff Data Scientist on the Experimentation Platform at Netflix, where he has spent the last four years specializing in computational causal inference. His work involves the development of novel statistical models and the integration of these models into scalable software solutions. Prior to Netflix, Simon was at Meta. Originally from Sweden, he holds a Ph.D. in Political Economy from Stanford University.