Pipelines

The Pipelines menu is used to define a series of steps required to build, train, and evaluate machine learning models. On the Cloudeka Deka MLOps Service Portal page, select the Pipelines menu.

The Pipelines menu provides three main features:

1. Upload Pipeline This feature allows you to upload a new pipeline definition to the Cloudeka Deka MLOps Service Portal environment. A pipeline definition is typically a YAML or JSON file that describes a workflow, including the sequence of steps, components, and parameters required for execution. By using the Upload Pipeline option, you can introduce new workflows to automate machine learning processes. Once uploaded, the pipeline can be accessed, edited, and executed as needed.

2. Refresh The Refresh option is used to update the list of pipelines displayed in the menu. This is particularly useful when changes have been made to pipelines, such as new uploads or modifications to existing pipelines. By clicking Refresh, you ensure that the displayed information is up to date and reflects the latest available pipeline versions in the Cloudeka Deka MLOps Service Portal environment.

3. Delete This feature allows you to permanently remove a selected pipeline from the system. Deleting a pipeline is useful when it is no longer needed or has become obsolete. Select the pipeline you want to delete and confirm the action to remove it from the list of available pipelines. Once deleted, the pipeline cannot be recovered unless it is uploaded or recreated again.

In the Pipelines menu, there are two types of pipelines: Private and Shared. Each type serves a different purpose and is designed to support various collaboration and access requirements within a team or organization.

a. Private Private pipelines are accessible only to the user who created them. This type of pipeline is suitable for individual experimentation and development, allowing users to work on workflows without interference from others. It also helps ensure that sensitive data or proprietary algorithms remain confidential. Once the pipeline is ready, it can be shared with other users if required.

b. Shared Shared pipelines are accessible to multiple users within the Kubeflow environment. This type of pipeline is useful for collaborative projects where team members work together on the same workflows. By sharing pipelines, users can contribute to the development process, leverage each other's expertise, and maintain consistency across different stages of a project. Shared pipelines also support better communication and collaboration when managing complex machine learning workflows involving multiple stakeholders.