Workspaces¶
A workspace is the top-level logical boundary in AIDF. It defines both the cloud account/region and the environment stage (Dev, QA, Staging, Prod). All projects, resources, and AI services live inside a workspace.
When to Create Workspaces¶
- One workspace per business unit (e.g., MUFG GCMD team, Legal team, Marketing)
- One workspace per country/region (e.g., Singapore, International)
- One workspace per major AI initiative (e.g., Customer Insight Generation)
Creating a Workspace¶
- Navigate to Workspaces tab → click
+Workspace - Choose provider: Amazon Web Services, Microsoft Azure, Google Cloud Platform, NVIDIA NGC
- Enter Identity Details: Admin user email (triggers approval workflow — only authorized personnel can create workspaces)
- Workspace Details: Name, description, cloud account, region
- Add Environment(s): Must add at least one. Choose type (Dev, QA, Stage, Prod) and size (Small/Medium/Large)
- Small: Light development, minimal data, low-end Kubernetes pod
- Medium: QA/staging, mirrors production performance without full scale
- Large: Production, high-throughput, multi-region HA clusters
- Submit for provisioning
Dynamic resource allocation
Creating an environment defines a quota (max GPU units, CPU cores, memory) without committing physical hardware. Resources are allocated from the shared pool only when a workload actively runs — optimizing cost by paying only for active computation.
Workspace Lifecycle¶
- Edit: Change description, add/modify environments
- Delete: Automatically deallocates ALL child projects, infrastructure (clusters, VMs, storage), and Kubernetes namespaces — preventing "zombie" resource costs