Projects & Workspaces

Projects are the top-level organizational unit in Strongly.AI. Each project groups together workspaces, data volumes, jobs, and collaborators into a single managed environment with persistent storage, access control, and activity tracking.

Core Concepts

Concept	Description
Project	Container for workspaces, volumes, jobs, and collaborators with shared settings
Workspace	Cloud-based development environment (JupyterHub, VS Code, or custom)
Data Volume	Persistent S3-backed storage with versioning, encryption, and backup
Job	Automated script, notebook, or command execution with scheduling
Collaborator	Team member with editor or viewer access to the project

Project Lifecycle

Create --> Active --> Archive --> Restore (or Delete)

Projects have three statuses:

• Active - Normal working state, all features available
• Paused - Temporarily suspended (no running workspaces or jobs)
• Archived - Read-only state; must be restored before making changes

Creating a Project

Step 1: Navigate to Projects

Click Projects in the main navigation sidebar, then click Create Project.

Step 2: Configure Project Details

Field	Required	Description
Name	Yes	Unique project name within your organization (1-100 characters)
Description	Yes	Purpose and goals (up to 500 characters)
Tags	No	Categorization labels (e.g., machine-learning, production)

Step 3: Automatic Setup

When you create a project, the platform automatically provisions:

1. Default Data Volume -- 50GB S3-backed storage with AES256 encryption, versioning enabled, and daily backups
2. Project README -- Markdown documentation template pre-filled with your project name and description
3. Default Environment -- JupyterHub configuration with 2 CPU, 8GB memory, 50GB disk
4. Activity Tracking -- All project actions are logged for audit purposes
5. Auto-Stop Settings -- Workspaces auto-stop after 30 minutes of inactivity by default

Project Categories

Projects can be categorized for organization:

Category	Use Case
machine-learning	ML model development and training
data-analysis	Data exploration and analytics
development	General software development (default)
research	Research and experimentation

Project Visibility

Level	Access
Private	Only owner and explicitly shared collaborators
Organization	All members in the same organization can access
Global	All platform users can access (Enterprise mode only)

Managing Projects

Updating Project Settings

Owners and editors can update project settings including name, description, tags, category, visibility, environment configuration, and notification preferences. Protected fields (owner, creation date, default volume, organization) cannot be changed.

Archiving a Project

Archiving puts a project into read-only mode. Before archiving:

• All workspaces must be stopped
• All running jobs must complete

To archive, navigate to the project and click Archive in settings. The archive timestamp is recorded for audit purposes.

Restoring an Archived Project

Click Restore on an archived project to return it to active status. All data volumes and settings are preserved.

Deleting a Project

Deletion is permanent. You have two options for data volumes:

• Keep Volumes -- Project volumes are converted to shared volumes, accessible from other projects
• Delete Volumes -- Project volumes and their data are permanently removed

A project with running workspaces cannot be deleted. Stop all workspaces first.

Collaborators

Adding a Collaborator

1. Navigate to the project's Settings tab
2. Click Add Collaborator
3. Enter the collaborator's email address
4. Select a role: Editor or Viewer
5. Click Add

The collaborator must already have an account on the platform. In multi-tenant (SaaS) mode, collaborators must be in the same organization.

Collaborator Roles

Role	Permissions
Owner	Full access: create, update, delete, manage collaborators, manage workspaces and volumes
Editor	Read and write access to all project resources; cannot delete project or manage collaborators
Viewer	Read-only access to project data, volumes, and job results

Updating Collaborator Roles

Owners can change a collaborator's role between editor and viewer at any time from the project settings.

Removing a Collaborator

Owners can remove any collaborator except themselves. Removing a collaborator immediately revokes their access to the project and all its resources.

Data Volumes

Data volumes provide persistent, S3-backed storage with enterprise features including versioning, encryption, automatic backups, and cross-project sharing.

Volume Types

Type	Scope	Description
Project	Single project	Default type; tied to a specific project lifecycle
Shared	Multiple projects	Can be attached to and accessed from multiple projects

Default Data Volume

Every project automatically receives a default data volume with:

• 50GB storage capacity
• AES256 encryption enabled
• Versioning enabled (up to 10 versions)
• Daily backups with 30-day retention
• Auto-sync on workspace stop
• Tagged with default and project-data

The default volume cannot be detached from its project.

Creating Additional Volumes

1. Navigate to the project's Data Volumes tab
2. Click Create Volume
3. Configure:

Field	Required	Options
Label	Yes	Descriptive name (1-100 characters)
Description	No	Purpose of the volume
Size (GB)	Yes	Storage capacity
Storage Class	No	standard (default), infrequent_access, glacier
Versioning	No	Enable version history
Backup	No	Enable automatic backups with frequency setting
Tags	No	Categorization labels

Volume Storage Classes

Class	Performance	Cost	Use Case
Standard	High throughput	Standard	Active development, frequently accessed data
Infrequent Access	Lower throughput	Reduced	Archival data accessed occasionally
Glacier	Retrieval delay	Lowest	Long-term archival, compliance data

Volume Data Formats

Volumes track the format of stored data:

• CSV, Parquet, JSON -- Single format volumes
• Images -- Image datasets
• Mixed -- Multiple data types (default)

Attaching and Detaching Volumes

Attach a shared volume to a project:

1. Go to the project's Data Volumes tab
2. Click Attach Volume
3. Select from available shared volumes within your organization
4. The volume appears in the project's volume list

Detach a volume:

1. Select the volume
2. Click Detach
3. The volume remains as a shared volume, accessible from other projects

The default project volume cannot be detached.

Making a Volume Shared

To share a project volume with other projects:

1. Navigate to the volume
2. Click Make Shared
3. The volume type changes from "project" to "shared"
4. Other projects in the organization can now attach it

Volume Sync

Volumes support automatic synchronization between workspaces and S3 storage:

Frequency	Behavior
On Workspace Stop	Sync when any workspace using this volume stops
Hourly	Automatic sync every hour
Daily	Automatic sync once per day
Weekly	Automatic sync once per week

Sync status is tracked as: idle, syncing, or failed.

Volume Encryption

All volumes support encryption:

• AES256 -- Server-side encryption with Amazon S3-managed keys (default)
• aws:kms -- Server-side encryption with AWS KMS-managed keys (requires KMS key ID)

Volume Versioning

When versioning is enabled, each version snapshot records:

• Version ID and number
• Created timestamp and user
• Size in GB
• Changelog description
• S3 storage location
• Tags

The maximum number of versions is configurable (default: 10).

Deleting a Volume

Volumes can be deleted with an optional archive flag. When deleted:

1. The volume is detached from all projects
2. If archive: true, the volume data is archived to S3 before deletion
3. The volume record is removed from the database

Three Volume Types in Workspaces

When working inside a workspace, you have access to three types of storage:

Volume Type	Mount Path	Lifecycle	Scope
Workspace Volume	/workspace	Persists through stop/start; deleted with workspace	Single workspace
Project Volume	/project/<name>	Persists through workspace deletion; deleted with project	All workspaces in project
Shared Volume	/shared/<name>	Persists until explicitly deleted	Multiple projects

Using Volumes in Code

import pandas as pd

# Workspace volume -- personal to this workspace
results_df.to_csv('/workspace/experiments/results.csv')

# Project volume -- shared across all workspaces in project
train_df = pd.read_csv('/project/training-data/train.csv')

# Shared volume -- organization-wide access
reference_df = pd.read_parquet('/shared/common-datasets/reference.parquet')

Important

Only files within /workspace, /project/*, or /shared/* persist. System-level installations (e.g., pip install, apt-get install) do not survive workspace restarts.

To persist Python packages, use a virtual environment:

python -m venv /workspace/venv
source /workspace/venv/bin/activate
pip install your-packages

Workspaces

Workspaces are cloud-based development environments running inside your project. Each workspace is a Kubernetes pod with its own compute resources, storage, and network configuration.

Development Environments

Environment	Image	Port	Best For
JupyterHub	jupyter/datascience-notebook	8888	Data science, ML, interactive notebooks, Python/R
VS Code	codercom/code-server	8080	Full-stack development, debugging, version control
Custom	User-provided	Configurable	Specialized development environments

Creating a Workspace

1. Navigate to your project
2. Click Create Workspace
3. Configure the workspace:

Field	Required	Description
Name	Yes	Workspace display name
Description	Yes	Purpose of this workspace
Environment	Yes	JupyterHub, VS Code, or Custom
Environment ID	Yes	Pre-configured environment template
CPU	Yes	CPU allocation (e.g., 1, 2, 4)
Memory	Yes	RAM allocation (e.g., 2GB, 8GB, 16GB)
Disk	Yes	Workspace volume size (e.g., 20GB, 50GB)
GPU	No	GPU allocation (e.g., 1 for NVIDIA T4)
Workspace Volume Size	No	Override default workspace storage (default: 20GB)

Connecting Services to Workspaces

Workspaces can be connected to platform services via the STRONGLY_SERVICES environment variable:

Service Type	Description
Add-ons	Managed databases (PostgreSQL, MongoDB, Redis, etc.)
Data Sources	External database connections
AI Gateways	AI model endpoints (OpenAI, Anthropic, vLLM, etc.)
Workflows	Workflow execution endpoints

These are selected during workspace creation and injected as a JSON environment variable.

Workspace Lifecycle

Created --> Deploying --> Running --> Stopping --> Stopped
                                         |
                                     Error / Archived

• Start -- Launch the workspace pod with configured resources
• Stop -- Gracefully stop the pod, triggering data volume sync if enabled
• Archive -- Delete the workspace and its workspace volume (project volumes are preserved)

Managing Workspaces

Start/Stop:

• Stop idle workspaces to save compute costs
• All volume data persists through stop/start cycles
• Auto-stop triggers after configurable idle time (default: 30 minutes)

Resource Upgrades:

• Stop the workspace before modifying resources
• Change CPU, memory, GPU, or disk allocation
• Restart with new configuration

Package Installation:

• JupyterHub: Use pip, conda, or apt-get in terminal
• VS Code: Install extensions from the marketplace
• Use virtual environments in /workspace/venv to persist packages

Multiple Workspaces per Project

Projects support multiple workspaces simultaneously. Each workspace:

• Has its own compute resources and environment configuration
• Shares access to all project data volumes
• Runs independently (starting/stopping one does not affect others)
• Has its own workspace volume at /workspace

Project Settings

Auto-Stop Configuration

Setting	Default	Description
Auto-Stop Enabled	true	Automatically stop idle workspaces
Auto-Stop Minutes	30	Minutes of inactivity before stopping

Data Volume Backup

Setting	Default	Description
Backup Enabled	true	Enable automatic backups
Backup Frequency	daily	Options: daily, weekly, manual
Retention Days	30	How long to keep backup copies

Notifications

Setting	Default	Description
Job Failure	true	Notify on job failures
Job Success	false	Notify on job completions
Workspace Auto-Stop	true	Notify when workspace auto-stops

Git Configuration

Projects support optional Git integration for version control:

Setting	Description
Repository URL	Git remote URL (e.g., git@github.com:user/repo.git)
Branch	Target branch (default: main)
SSH Key ID	Reference to user's SSH key for authentication
Sync Direction	pull, push, or both
Auto Sync	Automatically sync on schedule
Target Path	Path within volume to sync to (default: root)

Sync status is tracked with last sync timestamp, commit hash, and commit message.

Project Statistics

The platform tracks comprehensive project statistics:

Metric	Description
Total Jobs	Total number of jobs created
Successful Jobs	Jobs completed successfully
Failed Jobs	Jobs that failed
Running Jobs	Currently executing jobs
Total Storage (GB)	Combined storage across all volumes
Compute Hours Used	Total compute time consumed
Estimated Cost	Estimated cost of all resources

Activity Tracking

All project actions are logged for audit purposes:

Activity Type	Description
project_created	Project was created
project_accessed	Project was viewed
project_updated	Project settings were changed
workspace_started	Workspace was launched
workspace_stopped	Workspace was stopped
workspace_archived	Workspace was archived
job_started	Job execution began
job_completed	Job execution finished
volume_created	New volume was created
volume_updated	Volume settings were changed
volume_attached	Volume was attached to project
volume_detached	Volume was detached from project
collaborator_added	New collaborator was added
collaborator_removed	Collaborator was removed

Jobs

Jobs enable automated execution of scripts, notebooks, and commands within your project.

Job Execution Types

Type	Description	Configuration
Script	Execute a Python or shell script	Path, entrypoint, arguments, working directory
Notebook	Run a Jupyter notebook	Path, parameters
Command	Run an arbitrary command	Command string, arguments

Job Scheduling

Jobs can be triggered on various schedules:

Schedule Type	Description	Example
Cron	Standard cron expression	0 0 * * * (daily at midnight)
Interval	Fixed time interval	Every 15 minutes
Event	Triggered by platform events	Data volume updated, webhook, file change

Job States

State	Description
Active	Job is enabled and will run on schedule
Paused	Job schedule is suspended
Disabled	Job is deactivated

Job Environment

Each job runs in a containerized environment with:

• Custom Docker image
• Environment variables
• Secrets (injected from secure storage)
• Configurable CPU, memory, and optional GPU resources

GPU Options for Jobs

GPU Type	Description
nvidia-t4	NVIDIA T4 -- Good for inference
nvidia-a100	NVIDIA A100 -- High-performance training
nvidia-v100	NVIDIA V100 -- General ML workloads
nvidia-k80	NVIDIA K80 -- Cost-effective option

Access Control and Multi-Tenancy

Projects use the platform's multi-tenancy security model:

1. SaaS Mode -- Users can only access projects within their organization
2. Owner Access -- Project owner has full control
3. Admin Access -- Platform admins can access all projects
4. Collaborator Access -- Based on assigned role (editor/viewer)
5. Organization Visibility -- Organization members can access if visibility is set to "organization"
6. Global Visibility -- All users can access (Enterprise mode only)

Name uniqueness is enforced within each organization (excluding archived projects).

API Reference

Project Methods

Method	Parameters	Description
projects.create	{ name, description, tags? }	Create a new project
projects.update	projectId, updates	Update project settings
projects.delete	projectId, { archiveData?, deleteVolumes? }	Delete a project
projects.archive	projectId	Archive a project
projects.restore	projectId	Restore an archived project
projects.list	(none)	List all accessible projects
projects.getStats	projectId	Get project statistics
projects.trackAccess	projectId	Record project access
projects.getRunningJobsCount	projectId	Count running jobs

Collaborator Methods

Method	Parameters	Description
projects.addCollaborator	projectId, email, role, userId?	Add a collaborator
projects.removeCollaborator	projectId, userId	Remove a collaborator
projects.updateCollaboratorRole	projectId, userId, role	Change collaborator role

Workspace Methods

Method	Parameters	Description
projects.createWorkspace	projectId, options	Create a new workspace
projects.launchWorkspace	projectId, options?	Launch legacy single workspace
projects.stopWorkspace	projectId	Stop a running workspace
projects.archiveWorkspace	projectId	Delete workspace and its volume

Volume Methods

Method	Parameters	Description
volumes.create	{ projectId, label, description?, sizeGB, ... }	Create a new volume
volumes.sync	volumeId	Trigger volume sync
volumes.delete	volumeId, { archive? }	Delete a volume
volumes.makeShared	volumeId	Convert to shared volume
projects.attachVolume	projectId, volumeId	Attach volume to project
projects.detachVolume	projectId, volumeId	Detach volume from project
dataVolumes.getByIds	volumeIds[]	Get volumes by their IDs

tip

Start with a project closest to your needs. Use tags and categories to organize projects by team, department, or purpose. Take advantage of shared volumes to avoid duplicating large datasets across projects.