A QCon London talk shows how eBPF kernel hooks can filter prompts, swap models, and enforce token limits on AI traffic in Kubernetes without touching app code.
Continue to AI University →