Speaking

Ever wondered why OpenAI, Anthropic, or Google never send you those dreaded 'We'll be down for maintenance at 2 AM' emails, even though they're constantly pushing updates? The secret isn't magic — it's how they deploy. In this talk, we'll demystify zero downtime deployments by exploring the strategies used at scale: rolling updates, blue-green, canary releases, and feature flags. We'll strip away the buzzwords and show you how cloud-native tools like Kubernetes make them possible in practice. To keep things real, I'll also walk you through a live demo app — an AI-powered chat interface that pulls open-source models on the fly. You'll see how we can ship new versions without interrupting conversations, proving that zero downtime isn't reserved just for the big tech companies. If you're tired of 'system maintenance' emails and want to learn how to deliver reliable, continuous updates without drama, this session is for you.

AI apps don't sleep—and neither should your deployment strategy. In this talk we'll explore how to roll out updates to AI-powered services without a single second of downtime. From rolling updates and blue-green deployments to GPU-aware auto-scaling and live model hot-swaps, you'll see how cloud-native patterns keep even large language models online while you ship new features. A live demo will show an open-source AI chat app pulling new models in real time—proving that zero downtime isn't just for the big players.