The Invisible Intelligence Layer: Engineering Real-Time AI Systems That Serve Billions Without Breaking

Tejas Pravinbhai Patel, Senior Software Development Engineer at Amazon Retail, specializes in architecting large-scale distributed AI systems designed for real-time intelligence and global-scale applications. His work includes the Rufus real-time intelligence platform, serving hundreds of millions of customers worldwide. A Conference Senior Member with more than 20 peer-reviewed publications as a principal author, Tejas has also delivered keynotes at international academic and technology forums. His research and leadership in autonomous AI agents have earned recognition including a Best Paper Award, while his professional contributions include serving as ACM Irving Chapter Chair and SCRS Distinguished Fellow.

In his session, “The Invisible Intelligence Layer: Engineering Real-Time AI Systems That Serve Billions Without Breaking,” Tejas will explore the architectural foundations required to operate AI systems at massive scale while maintaining performance, reliability, and security. Drawing from real-world production experience, the session will examine distributed inference optimization, adaptive ranking, federated learning, and graceful degradation strategies, alongside the extension of Zero Trust principles into AI governance. Attendees will gain practical insights into the engineering trade-offs, failure modes, and architectural decisions that determine whether real-time AI infrastructure can remain resilient as demand and complexity grow.