Skip to main content
To work with NVIDIA Blueprint for virtual screening in Nebius AI Cloud, you can deploy it as an application in the web console.

​
Prerequisites

  1. Make sure you are in a group that has at least the editor role within your tenant; for example, the default editors group. You can check this in the Administration → IAM section of the web console.
  2. Make sure that you have enough quota for GPUs. The Blueprint requires a minimum of four GPUs to run.
  3. Create a Managed Service for Kubernetes® cluster that meets the following requirements:

​
How to deploy

  1. In the web console, go to https://mintcdn.com/nebius-ai-cloud/1Ha0sWR6e1mnIaHS/_assets/sidebar/applications.svg?fit=max&auto=format&n=1Ha0sWR6e1mnIaHS&q=85&s=06329add2f560a2a83d6c136ca5dfc9b Applications.
  2. Find NVIDIA Blueprint for virtual screening by searching for it or browsing by category, then open the application page.
  3. Click Deploy on cluster to deploy using the Kubernetes option.
  4. Configure the application:
    1. In Application name and Namespace, enter unique names for your deployment.
    2. Enter your NGC API key. If you do not yet have an NGC key, see How to get an NGC Key.
    3. Under Disk size, specify a value between 2000 Gi and 10000 Gi. At least 2000 Gi is required to download the models. Disks do not scale automatically, so enter a size that is sufficient for your needs.
    4. In JupyterHub admin password, click Generate to create a secure password, copy it and then select I saved the password securely.
    5. In JupyterHub accessibility, choose how the JupyterHub interface can be accessed.
  5. Click Deploy application.
  6. Wait until the application status becomes Running. The deployment might take up to three hours.
When deployment completes, a dialog with connection instructions opens. Follow it to get the JupyterHub URL or IP address depending on the selected access method.

​
How to get an NGC Key

If you do not yet have an NGC key, you can get one by the following steps:
  1. Go to NVIDIA NGC website.
  2. Enter your email, password and other necessary credentials to create an account.
  3. Once registered, go to build.nvidia.com and click Generate API key. Follow the steps and copy the API key and paste it in the deployment configuration.

​
How to access the Blueprint via JupyterHub

After the application is deployed, you can access it via JupyterHub by IP or by using port forwarding.

​
Access by IP

  1. Install and configure kubectl.
  2. Run the following command to get the external IP address: 
    kubectl describe svc proxy-public -n <namespace> | grep "LoadBalancer Ingress:"
  3. In your browser, go to http://<ip> and log in as admin. Use the password that you saved during deployment.

​
Access by port forwarding

  1. Install and configure kubectl.
  2. Set up port forwarding: 
    kubectl --namespace <namespace> port-forward service/proxy-public 8080:http
  3. Open http://localhost:8080 in your browser. On the login screen, enter admin as the username and use the password that you saved during deployment.

​
Managing users

To create and manage users in JupyterHub, do the following:
  1. In the JupyterHub top menu, select File → Hub Control Panel.
  2. Click Admin → Add Users.
  3. Enter usernames, one per line. To grant a user admin rights, select the Admin checkbox.
  4. Click Add Users.
New users create their own passwords the first time they sign in.

​
How to delete the Blueprint

  1. In the web console, go to https://mintcdn.com/nebius-ai-cloud/1Ha0sWR6e1mnIaHS/_assets/sidebar/applications.svg?fit=max&auto=format&n=1Ha0sWR6e1mnIaHS&q=85&s=06329add2f560a2a83d6c136ca5dfc9b Applications.
  2. Find the deployed Blueprint in the list and open it.
  3. In the application, go to the Settings tab.
  4. Under Delete application, click Delete.
  5. Confirm the deletion.

​
What’s next

When the application is running, you can access it via JupyterHub to run the sample notebook and experiment with the Blueprint.