Charmed Kubeflow documentation

Charmed Kubeflow (CKF) is an open-source, end-to-end, production-ready MLOps platform on top of cloud-native technologies.

It translates Machine Learning (ML) steps into complete workflows, including training, tuning, and shipping of ML models. It enables automation of workflows, increases quality of models, and simplifies deployment of ML workloads into production in a reliable way.

CKF meets the need of building ML applications in a structured and consistent manner while contributing to higher productivity and better collaboration within teams.

It is intended for data scientists and ML engineers, providing an advanced toolkit to organise and scale their work.


In this documentation

Tutorial
Get started - a hands-on introduction to CKF for newcomers
How-to guides
Step-by-step guides covering key operations and common tasks with CKF
Explanation
Discussion and clarification of key topics
Reference
Technical information, including specifications, APIs, settings and configuration

Project and community

Charmed Kubeflow is a member of the Ubuntu family. It’s an open-source project that welcomes community contributions, suggestions, fixes and constructive feedback.

Navigation

Navigation
Level Path Navlink
1 / Home
1 tutorial Tutorial
2 get-started Get started
2 build-your-first-ml-model Build your first ML model
1 how-to How to
2 install Install
3 install-general Install
3 install-in-an-airgapped-environment Install in an air-gapped environment
3 install-behind-a-web-proxy Install behind a web proxy
3 install-on-nvidia-dgx Install on NVIDIA DGX
3 install-on-aks Install on AKS
3 install-on-eks Install on EKS
3 install-on-gke Install on GKE
3 install-using-terraform Install using Terraform
2 manage Manage
3 upgrade Upgrade
4 upgrade-18-19 Upgrade from 1.8 to 1.9
4 upgrade-17-18 Upgrade from 1.7 to 1.8
3 uninstall Uninstall
3 troubleshoot Troubleshoot
3 backup Backup control plane
3 restore Restore control plane
3 integrate-with-cos Integrate with COS
3 integrate-with-minio Integrate with MinIO
3 integrate-with-mlflow Integrate with MLflow
3 manage-profiles Manage profiles
3 configure-high-availability-for-istio-gateway Configure High Availability for Istio Gateway
3 configure-the-kubeflow-notebook-creation-page Configure the Kubeflow Notebook creation page
3 deploy-nvidia-nim-on-charmed-kubeflow Deploy NVIDIA NIM
3 enable-https-on-charmed-kubeflow Enable HTTPS
3 enable-istio-cni-plugin Enable Istio CNI plugin
2 use Use
3 accelerated-ml-experiments-on-microk8s-with-inaccel-fpga-operator-and-kubeflow-katib Accelerated ML experiments on MicroK8s with InAccel FPGA Operator and Kubeflow Katib
3 customise-link-configuration-on-the-kubeflow-dashboard Customise link configuration
3 launch-nvidia-ngc-notebooks Launch NVIDIA NGC notebooks
3 serve-a-model-using-triton-inference-server Serve a model using Triton Inference Server
2 integrate-with Integrate with
3 integrate-with-identity-providers Identity providers
3 integrate-with-azure-blob-storage Azure Blob Storage
3 integrate-with-azure-spot-virtual-machines Azure spot virtual machines
1 reference Reference
2 release-notes Release notes
3 release-notes-1-9 Charmed Kubeflow 1.9
3 release-notes-1-8 Charmed Kubeflow 1.8
2 supported-versions Supported versions
2 monitoring Monitoring
3 prometheus-metrics Prometheus metrics
3 prometheus-alerts Prometheus alerts
3 grafana-dashboards Grafana dashboards
3 loki-logs Loki logs
2 kubeflow-bundle Kubeflow bundle
1 explanation Explanation
2 security Security
3 authentication Authentication
3 authorisation Authorisation
3 cryptography Cryptography
2 charmed-vs-upstream Charmed vs. upstream
2 mlops-tools MLOps tools
contributing-docs Contributing to docs
create-eks-cluster-for-mlops Setup EKS before installing CKF
create-aks-cluster-for-mlops Setup AKS before installing CKF
get-started-with-managed-kubeflow-on-azure Get started with Managed Kubeflow on Azure