Onyxia: A User-Centric Interface for Data Scientists in the Cloud Age
summary
Claims a concrete replacement for JupyterHub by running on Kubernetes and Helm and providing per-user Kubernetes namespace with S3 bucket and Vault directory per user, GUI buttons to launch a preconfigured S3-enabled Jupyter or VS Code, and support for Postgres, MLflow and custom services as Helm packages. Demonstrations show deployment on upstream OpenStack, easy forks and customizations, GPU scheduling tweaks, and a running footprint of 50+ instances worldwide with Helm chart packages that actually work.
AI-written summary · about
chapters (8)
- 0:00Onyxia introduction
- 3:59Cloud benefits
- 7:25Onyxia demo
- 10:13INSEE innovation
- 13:48Teaching with Onyxia
- 18:07Oceanography Digital Twin
- 23:46Customizing for oceanography
- 27:13Collaboration & future
related talks
37mJupyterHub from the Ground Up with Kubernetes
34mMoving notebooks into the cloud: challenges and lessons learned
39mEnabling Data centric Teams
30mAzimuth - self service cloud platforms for reducing time to science
24mJupyter Notebooks for Ops
34mEmpower Developer Experience
48mSponsored Session: Your Next Workstation Is In The Cloud
28mA Walk Through the Kubernetes UI Landscape