Скачать или смотреть AI on Kubernetes Considerations: Tools and Techniques

Running AI—especially GenAI and LLM workloads—on Kubernetes isn’t just about spinning up containers. It demands careful planning across infrastructure, scaling, observability, and resource management. In this video, we dive into what to watch out for and which tools can help.

What you’ll learn:
Why event-driven infrastructure (Knative, KEDA) matters for prompt-based & streaming workloads
How to handle autoscaling, latency, and resource intensity for model serving and inference
Observability, prompt & model drift, and best practices for maintaining consistency over time
Tools & components like VPA, KServe, Kueue, Kubeflow Pipelines, GPU scheduling, etc.

If you're designing or operating AI systems on Kubernetes, this video is for you. Don’t forget to like, comment your experiences, and subscribe for more cloud-native insights.