AI & the future of Kubernetes with Abdel Sghiouar | Weave Intelligence скачать в хорошем качестве

AI & the future of Kubernetes with Abdel Sghiouar | Weave Intelligence

Kubernetes isn’t disappearing - it’s becoming invisible. And AI is the reason that matters. In this episode, Sam sits down with Abdel Sghiouar (host of the Kubernetes Podcast and long-time Kubernetes expert) to unpack how AI is reshaping the platform engineering landscape - without replacing the fundamentals. They start with a surprising data point: Kubernetes dropped from ~70–80% to ~44% as the “main focus” in platform engineering surveys - not because teams stopped using it, but because Kubernetes is increasingly assumed as the default substrate while new platform domains (FinOps, data, AI, security) explode around it. From there, Abdel goes deep on what’s changing inside Kubernetes itself: the community’s shift toward better AI workload support, new working groups focused on accelerators and inference, and why capabilities like Dynamic Resource Allocation (DRA) matter far beyond GPUs. They also explore the cultural side: new hybrid personas (software engineer + “agent builder”), why product managers are suddenly learning Kubernetes, and what “platform capabilities” look like when every app wants guardrails, prompt filtering, and secure LLM access. Finally, the conversation gets practical: what junior engineers should do right now, why fundamentals still beat hype, and what Abdel thinks the next wave of “agentic platforms” will actually be built from - gateways, sandboxes, and emerging agent standards (MCP, agent-to-agent protocols, and more). In this episode: Why Kubernetes is becoming the “default layer” you don’t talk about — but still rely on What the Kubernetes project is doing for AI workloads: accelerators, inference, and new working groups DRA (Dynamic Resource Allocation) explained — and why it matters beyond AI The new personas AI is creating (and why roles are blurring fast) Guardrails as a platform capability: prompt filtering, policy, and security patterns “Agentic platforms” demystified: what’s actually new vs. “just an app calling an LLM” 2026–2027 focus areas: agent gateways, sandboxing, and agent protocol standards Advice for juniors: leverage AI, learn fundamentals, and build faster with better tools Global maturity: why some regions leapfrog — and why opportunity is still everywhere 💬 “When you hear ‘platform,’ you should probably assume there’s Kubernetes somewhere underneath.”