SUSE AI Factory
This guide walks you through creating custom blueprints for SUSE AI Factory using the command line. Learn how to quickly adapt an existing blueprint and how to build a full multi-component stack from scratch, supporting both CPU (Ollama) and NVIDIA GPU (vLLM) inference.
The instructions assume an operational Kubernetes cluster with Rancher Prime and SUSE AI Factory installed (including NVIDIA GPU support if using vLLM), and that you maintain your blueprint files in a Git repository.
Publication Date: 2026-11-02
- 1 What is a blueprint?
- 2 What are the prerequisites?
- 3 Quick start: copy an existing blueprint
- 4 What is the blueprint directory structure?
- 5 How do I create a blueprint CR for inference with Ollama?
- 6 How do I create a blueprint CR for GPU inference with vLLM?
- 7 How do I manage additional Helm repositories?
- 8 How do I document a blueprint?
- 9 How do I publish and run a blueprint?
- 10 How do I troubleshoot blueprint deployment issues?
- A Blueprint templates and reference architectures