Workspace
A Workspace is a versatile container within an account, designed to model and segregate specific use cases, domain areas, or team structures. It contains multiple projects grouped around a common operational theme or purpose. Workspaces also provide distinct boundaries for access control, enabling different sets of users, teams, or sub-organizations to independently manage their workspaces.
¶ Key characteristics of a Workspace:
-
Each workspace is identified by a unique identifier within the account.
-
Workspaces contain multiple projects, grouped based on a shared operational theme or purpose.
-
Workspaces provide shared infrastructure and configurations, including cloud storage connections and encryption policies, which are inherited by all projects within.
-
It adapts flexibly to various client scenarios, such as intent-based use cases (e.g., synthetic data creation, AI/ML training), organizational structures (e.g., customer support, autonomous vehicles), or labeling types (e.g., text labeling, image labeling).
Below are some example scenarios and how workspaces/projects can be defined for them.
| Scenario Type | Example Workspace | Example Project |
|---|---|---|
| Purpose/Intent Based | Synthetic Data Creation | Training data creation for Customer Service Chatbot |
| Demonstration dataset (prompt-response pair) | ||
| Comparison dataset (prompt- winning response - losing response) | ||
| AI/ML Training Workspace | Food Items Labeling | |
| Food Calories Labeling | ||
| Food Images Inbound Workspace | Food Detection Labeling | |
| Food Types Labeling | ||
| Orgs/Team Based | Customer Support | Fraud Detection Labeling |
| Automation ML Training | ||
| Autonomous Vehicles | Car Color Labeling | |
| Lidar Labeling | ||
| Labeling Type Based | Text Labeling Workspace | Intent Detection Labeling |
| Image Labeling Workspace | Consensus Based Image Labeling | |
| On-demand Labeling |