All Products
Document Center

Platform For AI:Use workspaces to manage permissions

Last Updated:Feb 08, 2024

Workspace is a key concept in Platform for AI (PAI). In a workspace, your organization or team can manage computing resources, user permissions, and AI assets in a centralized manner and achieve seamless collaboration at every stage of AI development.

Workspace features

Workspaces provide the following features:

  • Computing resource management: Your team can manage all computing resources for AI development on a single platform.

  • Permission management: You can have fine-grained control over the permissions of each member.

  • AI asset management: You can save and manage AI models and algorithms for further improvement and team collaboration.

  • Event monitoring and alerting: You can create an alerting rule to monitor the status of Deep Learning Containers (DLC) jobs or Machine Learning Designer pipeline jobs or trigger related events based on the approval status of model versions.

  • Resource scheduling: You can allocate resource quotas, determine the maximum runtime for jobs, and assign job priorities based on the components in use and the member roles. This ensures an optimal distribution of resources.

Member roles

PAI provides the following predefined roles for AI development and management:

  • Resource administrator: This role must be assigned to an Alibaba Cloud account or an authorized RAM user. Members with this role can add or delete workspaces and create Data Science Workshop (DSW) instances.

  • Workspace administrator/owner: Members with this role can manage members and public assets in the workspace.

  • Algorithm developer: Members with this role can manage their assets and public assets.

  • Algorithm O&M engineer: Members with this role can manage all training jobs in the workspace, including viewing the DLC jobs that are created by other members.

  • Labeling administrator: Members with this role can create datasets that are used for labeling and update their datasets.

  • Visitor: Members in this role can only view workspace information, such as members, resource groups, and pipelines.

  • MaxCompute developer: Members with this role can use MaxCompute to execute jobs that are submitted in the workspace.

For information about the permissions that are granted to each role or required by each module in PAI, log on to the PAI console and go to the Roles and Permissions page.

Workspace instructions

For information about how to create and manage workspaces and manage workspace notifications, see Workspace.

AI computing asset management

For information about how to manage AI computing assets in PAI, such as datasets, models, images, and jobs, see AI Computing Asset Management.