Kubeflow: ML on OpenShift

on YouTube ↗40m

" summary

Kubeflow on OpenShift aims to deliver portable, composable ML on Kubernetes and currently ships as v0.4 with ~23 repos, ~143 contributors and 1,000+ commits; core pieces include JupyterHub notebooks (curated TensorFlow images back to 1.4, current ~1.12), TFJob training and Kubeflow Pipelines (Argo-based) plus Ambassador ingress and SeldonIO for serving. Practical gotchas: many community images assume flexible UIDs and need cluster-admin or SELinux tweaks on OpenShift 3.11; workshop deployments require elevated rights to avoid port/UID failures.

AI-written summary · about

" chapters (9)

1. 0:00Start
2. 4:12Gpus
3. 13:54Community
4. 14:55Integrations
5. 20:55Notebooks
6. 25:22Argo
7. 30:44The Challenges with Openshift
8. 31:58Operator Model
9. 35:57Demo

" related talks

• 14mKubeflow: Where ML Meets KubernetesSuresh Peiris · KCD Sri Lanka 2023
• 30mKubeflow, Your ML Tools on KubernetesLaura Morillo, Iván Cañaveral · Commit Conf 2018
• 32mKubeflow Deep DiveDavid Aronchick, Jeremy Lewi · KubeCon + CloudNativeCon Europe 2018
• 37mTowards Kubeflow 1.0, Bringing a Cloud Native Platform For ML to KubernetesDavid Aronchick · KubeCon + CloudNativeCon Europe 2019
• 46mKubeflow, TensorFlow on KubernetesIsrael Olalla · Commit Conf 2018
• 17mEmpowering the Future of ML: Scaling the Kubeflow Community in 2026Amita Sharma, Gaurav Kamathe · DevConf.IN 2026
• 51mMachine Learning using Kubeflow and KubernetesArun Gupta · Devoxx Belgium 2019
• 31mIntroduction to Using and Use Cases of KubeFlowJonathan Gershater, Boris Lublinsky · Open Source Summit + Embedded Linux Conference North America 2019