Getting Started With Vantage

This tutorial walks you through the complete Vantage onboarding path, from creating an account to running a job and launching a notebook. By the end you will have a working organization, a connected cloud account, a live cluster, a submitted job, and an interactive notebook session.

Prerequisites

• A modern web browser
• A Google or GitHub account for OAuth sign-in
• Credentials for at least one cloud provider (AWS, Azure, or GCP), or access to an LXD / on-premises environment

Step 1: Sign up and create your organization

1. Open app.vantagecompute.ai and click Sign Up below the Continue button.
2. Click Continue with Google or Continue with GitHub to authenticate.
3. After signing in for the first time, the Welcome to Vantage modal appears. Fill in:

Field	Required	Notes
Organization Name	Yes	Displayed throughout the platform
Organization ID	Auto	Generated from the name. Lowercase, numbers, and dashes only. Cannot be changed after creation.
Logo URL	No	URL to an image used as your organization logo

4. Click Create Organization.
5. A 5-step onboarding tour highlights the key areas of the platform. Click Next to advance or Skip the Tour to go straight to the dashboard.

Step 2: Invite a teammate

1. Click Settings in the left sidebar, then select the Users tab.
2. Click New Invite in the upper-right corner.
3. Enter the invitee's email address and click Add Permission Group to assign at least one group (for example, Regular User).
4. Click Send invite request. The invitee receives an email with a link to join your organization.

For details on permission groups, teams, and SSO, see the IAM how-to guides and Teams how-to guides.

Step 3: Connect a cloud account

Cloud accounts link your external infrastructure to Vantage so the platform can provision clusters on your behalf.

1. Click the Settings icon (gear) in the bottom-left sidebar and select Cloud Accounts.

2. Click + Create Cloud Account and choose your provider (GCP, Azure, AWS, LXD, or On-Premises).

3. In the dialog that opens, enter an Account Name and fill in the provider-specific fields:

   • GCP / Azure / On-Premises -- enter a name and optional description, then click Create Cloud Account.
   • AWS -- for first-time setup, use Assisted Setup: click Access the AWS Stack Creation page, create the CloudFormation stack in AWS, wait for CREATE_COMPLETE, return to Vantage, and click Next. Alternatively, select Existing ARN Configuration and paste your IAM Role ARN.
   • LXD -- provide the Server URL, Client Certificate, and Client Key.

Your new account appears in the Cloud Accounts list and is now available when creating clusters.

For the full provider reference, see Compute providers.

Step 4: Create your first cluster

Clusters are the compute environments where jobs and notebooks run.

1. Click Clusters in the left sidebar and choose the cluster type tab (Slurm or Kubernetes).
2. Click + Prepare Cluster.
3. Fill in the Configure step:

Field	Required	Notes
Cluster Name	Yes	Max 27 characters, must be unique
Description	No	Max 255 characters
Cloud Account	Yes	Select the account you just created

4. Non-AWS providers -- click Create Cluster to submit with backend defaults.

   AWS -- click Continue, then select a Region, Head Node Machine Type, and SSH Key Name. For Slurm clusters, proceed to the Partitions step and configure at least one partition. Click Prepare Cluster to submit.

5. The cluster appears in the list with a "preparing" status. Wait for it to transition to "ready" (green badge).

For on-premises clusters (Multipass, Juju, Ansible, Terraform), see On-Premises clusters.

Step 5: Write a job script

1. Click Jobs in the left sidebar, then select Scripts.
2. Click Create Script, enter a Name, and click Create Script. You are redirected to the script detail page.
3. Click the edit icon next to the entrypoint file to open the built-in editor and paste a script:

#!/usr/bin/env bash
#SBATCH --job-name=my-first-job
#SBATCH --output=my-first-job.%j.out
#SBATCH --error=my-first-job.%j.err
#SBATCH --time=00:10:00
#SBATCH --ntasks=1
#SBATCH --cpus-per-task=2
#SBATCH --mem=512M

set -euo pipefail
echo "Job start: $(date)"
echo "Hello from $(hostname)"

A Slurm script has three sections: the shebang (#!/usr/bin/env bash), #SBATCH directives (resource requests), and the workload (commands to run).

4. Click Save.

Step 6: Submit your first job

1. From the script detail page, click Actions and select Submit Script. The Submit dialog opens with the script pre-selected.
2. Fill in the submission form:

Field	Required	Notes
Submission Name	Yes	Pre-filled from the script name
Cluster	Yes	Start typing to search
Partition	Yes	Populates after selecting a cluster

3. Click Submit. You are redirected to the Submissions page.
4. Watch the job progress through Pending, Running, and Completed (or Failed). Click the submission row to view logs and metrics.

Step 7: Launch a notebook

1. Click Workbench in the left sidebar, then click Sessions in the Workbench sidebar.
2. Click + New session.
3. Configure the session:

Field	Required	Notes
Session Name	Yes	Default: my-session
Preset	No	Pre-configured environment if available
Compute Pool	No	Defaults to your workspace pool
CPU	No	Default: 0.5
Memory	No	Default: 1Gi

4. Click Launch session. The session appears with a "Starting" status.
5. When the status changes to "Running", click Open to launch JupyterLab, VS Code, or RStudio in your browser.

What's next

You now have the complete foundation: an organization, a cloud account, a cluster, a submitted job, and a running notebook.

Explore the how-to guides to go deeper:

• Create a Slurm cluster or Create a Kubernetes cluster
• Create job templates for reusable parameterized scripts
• Submit a training job or Create a pipeline
• Deploy a model endpoint
• Work with storage (PVCs, NFS, CephFS)
• Set up license servers for commercial software
• Install the CLI for command-line automation