[[["容易理解","easyToUnderstand","thumb-up"],["確實解決了我的問題","solvedMyProblem","thumb-up"],["其他","otherUp","thumb-up"]],[["難以理解","hardToUnderstand","thumb-down"],["資訊或程式碼範例有誤","incorrectInformationOrSampleCode","thumb-down"],["缺少我需要的資訊/範例","missingTheInformationSamplesINeed","thumb-down"],["翻譯問題","translationIssue","thumb-down"],["其他","otherDown","thumb-down"]],["上次更新時間:2025-09-04 (世界標準時間)。"],[],[],null,["# Set up a project for a team\n\nThis page provides examples for how you might set up a project for\na team working with Vertex AI. This page assumes that you're\nalready familiar with Identity and Access Management (IAM) concepts such as policies,\nroles, permissions, and principals as described in\n[Vertex AI access control with IAM](/vertex-ai/docs/general/access-control)\nand [Concepts related to access\nmanagement](/iam/docs/overview#concepts_related_to_access_management).\n\nThese examples are intended to be used generally. Consider the team's\nspecific needs, and adjust how you set up the project accordingly.\n\nOverview\n--------\n\nVertex AI uses IAM to manage access to\nresources. When you plan access control for your resources, consider the\nfollowing:\n\n- You can manage access at the project level or resource level.\n Project-level access applies to all of the resources in that project.\n Access to a specific resource only applies to that resource.\n\n- You grant access by assigning IAM roles to principals.\n Predefined roles are available to make it easier to set up access,\n but custom roles are recommended because you create them, so you can\n limit their access to only the permissions that are required.\n\nTo learn more about access control, see [Vertex AI access control\nwith IAM](/vertex-ai/docs/general/access-control).\n\nSingle project with shared access to data and Vertex AI resources\n-----------------------------------------------------------------\n\nIn this example, a team shares a single project that contains their data\nand Vertex AI resources.\n\nYou might set up a project this way if the team's data, containers, and other\nVertex AI resources can be shared among all users of the project.\n\nYour project's IAM allow policy might look similar\nto the following: \n\n```json\n{\n \"version\": 1,\n \"etag\": \"BwWKmjvelug=\",\n \"bindings\": [\n {\n \"role\": \"roles/aiplatform.user\",\n \"members\": [\n \"user:USER1_EMAIL_ADDRESS\",\n \"user:USER2_EMAIL_ADDRESS\"\n ]\n },\n {\n \"role\": \"roles/storage.admin\",\n \"members\": [\n \"user:USER1_EMAIL_ADDRESS\",\n \"user:USER2_EMAIL_ADDRESS\"\n ]\n },\n {\n \"role\": \"roles/aiplatform.serviceAgent\",\n \"members\": [\n \"user:service-PROJECT_NUMBER@gcp-sa-aiplatform-cc.iam.gserviceaccount.com\"\n ]\n }\n ]\n}\n```\n| **Note:** The project number is a unique identifier for your project. See [Find the project name, number, and\n| ID](/resource-manager/docs/creating-managing-projects#identifying_projects).\n\nSetting up a project this way makes it easier for a team to collaborate\nto train models, debug code, deploy models, and observe endpoints.\nAll users see the same resources and can train with the same data.\nVertex AI resources operate within a single project, so you\ndon't need to grant access to resources outside of the project. Quota\nis shared across the team.\n\nTo set up access control for your team's project,\nsee [Manage access to projects, folders, and\norganizations](/iam/docs/granting-changing-revoking-access).\n\nSeparate data and Vertex AI resources\n-------------------------------------\n\nIn this example, the team's data is located in a project separate from\nVertex AI resources.\n\nYou might set up a project this way if:\n\n- The team's data is too difficult to move to the same project as your\n Vertex AI resources.\n\n- The team's data requires specific control over who can access it.\n\nIn these situations, we recommend creating a project for the data and a\nproject for Vertex AI resources. The team's developers\nshare the project that contains the Vertex AI resources.\nThey use the Vertex AI resources to access and process the\ndata stored in the other project. Data administrators grant the\nVertex AI resources access through service agents or\ncustom service accounts.\n\nFor example, you might grant the default Vertex AI service agents\naccess to a Cloud Storage bucket with an allow policy that looks like\nthe following: \n\n```json\n{\n \"version\": 1,\n \"etag\": \"BwWKmjvelug=\",\n \"bindings\": [\n {\n \"role\": \"roles/storage.objectViewer\",\n \"members\": [\n \"user:service-PROJECT_NUMBER@gcp-sa-aiplatform-cc.iam.gserviceaccount.com\",\n \"user:service-PROJECT_NUMBER@gcp-sa-aiplatform.iam.gserviceaccount.com\"\n ]\n }\n ]\n}\n```\n\nWhen possible, specify a service account to use as the resource identity when\nyou create Vertex AI resources, and use that service account\nto manage access control. This makes it easier to grant specific resources\naccess to the data and manage permissions over time.\n\nFor example, you might grant a service account access to\nBigQuery with a policy that looks like the following: \n\n```json\n{\n \"version\": 1,\n \"etag\": \"BwWKmjvelug=\",\n \"bindings\": [\n {\n \"role\": \"roles/bigquery.user\",\n \"members\": [\n \"user:SERVICE_ACCOUNT_NAME@PROJECT_NUMBER.iam.gserviceaccount.com\"\n ]\n }\n ]\n}\n```\n\nTo set up access control for service accounts, see\n[Manage access to service accounts](/iam/docs/manage-access-service-accounts).\n\nIn the project with the Vertex AI resources, administrators\ncan grant users access to the data by granting the Service Account User role\n(`roles/iam.serviceAccountUser`) on the specified service accounts.\n\nIsolate less trusted code in additional separate projects\n---------------------------------------------------------\n\n\nModels, prediction containers, and training containers are code. It's important to isolate\nless trusted code from sensitive models and data. Deploy endpoints and training stages\nin their own projects, use a dedicated service account with very limited permissions,\nand use VPC Service Controls to isolate them and reduce the impact of access granted to such\ncontainers and models.\n\nWhat's next\n-----------\n\n- To learn more about endpoint access control, see [Control access to\n Vertex AI endpoints](/vertex-ai/docs/general/endpoint-access-control).\n\n- To learn more about using a custom service account to control access\n to specific resources, see\n [Use a custom service account](/vertex-ai/docs/general/custom-service-account)."]]