Ollama Operator
A Kubernetes operator that runs Ollama-served large language models on a cluster, letting a user deploy and manage multiple models the way they would manage any other Kubernetes workload.
Ollama Operator installs CRDs on a Kubernetes cluster, and after that, creating a model resource is enough for the operator to fetch and load it, similar to using the Ollama CLI locally but reproduced with Kubernetes-native abstractions. It works on general Kubernetes clusters as well as smaller distributions such as K3s on a Raspberry Pi or TrueNAS SCALE, kind and minikube, needs no dependencies beyond Kubernetes itself, and lets a node share a pulled model image the same way container images are shared.
A companion kollama command-line tool makes deploying and exposing a model, for example with a NodePort service, a single command instead of writing Kubernetes manifests directly. Because it uses llama.cpp under Ollama, there is no need to separately manage a Python environment or CUDA drivers on the cluster.