The Model Context Protocol (MCP) has become the de facto standard for connecting Large Language Models to enterprise systems — but what works beautifully on localhost falls apart the moment you introduce Kubernetes, real traffic, and distributed state.
This session covers the three architectural realities every team hits in production: the migration from Server-Sent Events (SSE) to Streamable HTTP and why SSE’s persistent connection model is fundamentally incompatible with cloud-native deployments; Redis-backed distributed session management that survives pod restarts and autoscaling events; and a battle-tested Kubernetes deployment blueprint covering ingress configuration, horizontal pod autoscaling, and transport-layer security hardening.
You’ll leave with concrete patterns and hard-won lessons for running MCP where it actually matters — in production.

AI-Native Engineering
TBA
Thomas Pliakas
TBA

Stateless Lies and Distributed Truths: Running MCP at Scale on Kubernetes