Optimizing Video Playback at Scale: An End-to-End Approach with SODA and AZEEM

Abstract: Large-scale video streaming services are among the most resource-intensive applications on the modern Internet, serving tens of millions of concurrent users. Delivering a high quality of experience (QoE)—often defined by minimal rebuffering, fast startup times and sustained high bitrates—is paramount for user retention. In this talk, I will present two complementary approaches which deliver high QoE at scale: SODA a practical and theoretically sound client-side adaptive bitrate algorithm which optimizes quality selection through principled mathematical modeling and AZEEM, a server-side framework that leverages few-shot learning to dynamically optimize streaming parameters at scale. Both solutions are battle tested in production and I will describe design underpinnings as well as the empirical results from large-scale deployments. Finally, I’ll conclude with an overview of Amazon Prime Video’s science programs including focus areas and academic collaboration opportunities.

Bio: Zahaib Akhtar is a Senior Applied Scientist at Amazon Prime Video and serves as an Adjunct Assistant Professor in the Computer Science Department at North Carolina State University (NCSU). He earned his Ph.D. in Computer Science from the University of Southern California in 2019. At Amazon Prime Video he architects and deploys systems that enhance streaming experience for millions of users. His research investigates a range of areas such as streaming quality optimizations, performance observability systems and large-scale operations. His work regularly bridges the gap between academic research and industry applications by bringing state of the art techniques to production-scale video streaming challenges.

Website: https://zahaibakhtar.com/