Aneri Shah

Personalizing Customer Experience Using Large-Scale Distributed Systems and Event-Driven Architecture

Abstract:

Modern consumer platforms serving tens of millions of users face unprecedented challenges in delivering real-time, context-aware personalization with sub-second latency and 99.99% availability. Traditional monolithic architectures are fundamentally incapable of supporting billions of daily interactions while meeting these performance demands. This presentation explores how distributed systems principles, combined with event-driven architecture, enable organizations to build world-class personalization systems at internet scale. Drawing on hands-on experience at Amazon and Yahoo—where systems serve hundreds of millions of daily active users—the talk traces the full personalization lifecycle, from event ingestion and real-time processing to delivering tailored recommendations. We will cover eight core areas: defining personalization requirements at scale; architectural patterns for horizontal scalability; event-driven pipelines processing billions of interactions; recommendation systems using collaborative filtering and contextual bandits; consistent customer state management across microservices; real-time feature computation with sub-100ms latency; observability and resilience in distributed systems; and case studies demonstrating measurable impact, including 30–40% increases in engagement and 15–25% improvements in retention. Attendees will gain practical insights into architectural patterns, technology choices (including Kinesis, DynamoDB, Kafka, Lambda, and Flink), and operational strategies that power large-scale personalization. Key takeaways include understanding why events serve as the source of truth, how eventual consistency enables performance at scale, the importance of real-time feature computation, and how graceful degradation ensures reliability during system failures. This talk is designed for software architects, backend engineers, data engineers, and technical leaders interested in building scalable, personalized experiences for global platforms.

Profile:

 

Aneri Shah is a Global Engineering Leader at Amazon, where she leads engineering initiatives focused on large-scale distributed systems and cloud-based data platforms that power real-time consumer applications. She oversees the design and evolution of backend services, scalable APIs, and event-driven architectures that manage customer engagement and state across Amazon Music’s global platform serving more than 80 million users.  In her role, she has led major platform modernization initiatives, including a multi-year transformation that unified fragmented legacy library and engagement systems into a single distributed architecture built on AWS technologies such as DynamoDB, Kinesis, Lambda, and Glue, while successfully orchestrating large-scale migrations of historical customer data across multiple interconnected services. Her work has significantly improved system reliability, cross-device synchronization, and long-term platform scalability.

Beyond her engineering leadership, she actively mentors early career engineers and leads outreach initiatives such as Amazon Music’s “Girls Who Don’t Code Yet” program, introducing high school students to technology careers and encouraging greater participation of women in the tech industry. Through technical publications in medium.com, she shares insights on distributed systems, event-driven architectures, and scalable platform design, contributing to broader industry discussions on building resilient, evolving software systems.