What is ZenML Cloud?
ZenML Cloud is a SaaS backend for your team’s local or cloud-based ZenML pipelines. Your ML workflow won’t change - you still could run ZenML on your local machine, or use an integration into a processing backend, but your metadata and pipeline definition is stored centrally.
Compliance and governance won’t change - you are still in charge of where the data is stored, where processing happens and where models get deployed. But ZenML Cloud simplifies working together in a team:
- Centralized pipeline store
- Pipelines are persisted not only in the environment they are created, but also on ZenML Cloud to allow for discoverability and reusability throughout your organisation.
- Centralized Metadata Store
- Metadata (and the Artifact Store destination) is persisted centrally to allow for caching and comparability across your organization.
- Centrally managed backends
- Reference backends from your pipelines, without knowledge about their config or having permissions spread out across laptops.
- Share backends with Data Scientists without sharing access permissions.
- Hosted pipeline evaluation, comparison and statistics
- Get instant insights into your pipeline results through hosted evaluation notebooks, comparisons across pipelines and statistics per processing step.
- Managed pipeline invocation
- pipeline.run() sends the pipeline.yaml to our pipeline registry and launches the orchestration job on a preconfigured backend - 0 knowledge necessary for the invoking user
- Managed pipeline orchestration
- ZenML Cloud can orchestrate your pipelines - zero self-managed infrastructure on your end required. We launch pipeline steps in your processing environment, so you only pay for processing resources.
- Manage teams and projects in Organizations and Workspaces
- User permission management
- Project-specific configurations for experiments via Workspaces