Our experience of deploying Google’s Kubernetes ML toolkit on physical servers with multiple GPUs. The talk will cover all the related topics, from zero to hero:
- Choosing the hardware for deep learning.
- Installing, configuring, and optimizing the OS for deep learning.
- Organizing a simple but DL-friendly distributed file system.
- Installing Kubernetes and preparing it for Kubeflow.
- Installing Kubeflow and making the next steps.
- What’s MLOps and why you should hire one.