Skip to content

Get started

To manage or create notebook jobs, go to Workspaces, select a workspace from the list, then click Notebook Jobs.

This opens the Notebook Jobs page, which lists all jobs in the current workspace.

From here, you can:

  • Monitor job status
  • Access job details
  • Filter, review, or rerun jobs

Step 1. Draft your job

When you want to analyze a workspace's shared data, the first step is to create a job and write a query on one or more datasets.

Create a job

To create a new notebook job:

  1. Go to Workspaces Notebook Jobs and click New Job.
  2. Enter a name (max 50 characters) and optional description (max 150 characters).
  3. (Optional) Enable Request approval for reruns if you plan to run this job more than once.
  4. Click Save to create the job.

The new job is added to the Jobs page table with a Draft status.

The Notebook jobs list page

Write your query

Next, you can begin to define your job.

  1. In the Jobs table, click the job name to open the Job details page.
  2. In the Query tab, begin writing your logic using Python (PySpark) or PySpark SQL.
    • We recommend using a starter script, which includes the required structure for loading encrypted datasets and saving results.
    • See Write queries for details on using starter scripts and referencing datasets.
    • Note: Inserting a starter script replaces the current contents of the editor.

Job edits are saved automatically. You can return at any time to continue writing or test your logic using synthetic data.

While in draft, the job and its query are visible to all workspace members.

The Notebook job detail page

Test your job

While your job is still in draft, you can test it using synthetic data. This lets you fine-tune your logic and ensure the job runs as expected—without ever exposing real data.

To test your job:

  1. In the Query editor, complete your logic using Python (PySpark) or SQL with PySpark.

  2. At the top of the editor, click Job Actions, then select Test on Mock Data.

    Note

    The dataset you're querying must have test data available. Look for a clipboard icon next to the dataset name in the sidebar—hover to see whether test data is available.

  3. Open the Run History tab to view test results and logs.

You can continue editing your job as long as you haven’t submitted it for review or run. Once a job is under review, it becomes read-only—but you can cancel the review to return it to draft. After a job is run, you can no longer edit it. For more information on testing, see Use test data.

The Notebook job mock data

Step 2. Prepare for review

The steps in this section apply to members in a multiparty workspace. If you're in a single-party workspace, you can skip this section.

Submit your job for review

When your job is ready for approval, click Submit for Review.

  • The status of your job changes from Draft to Under Review.
  • Workspace members are notified and can begin reviewing.
  • Use the Job Status panel to view each member’s response.

Known limitation

Members receive a general status notification—not individual review requests. You may want to notify reviewers separately.

Revise and resubmit (if rejected)

If a reviewer rejects the job:

  1. Open the Job Status panel, locate the Declined row, and read the reviewer’s comment.
  2. On the Query tab, click Cancel Review to move the job back to Draft.
  3. Make your edits, then click Submit for Review again.

Review a job

To review a notebook job:

  1. On the Jobs list page, select the job name, or click View job in your Notifications.
  2. In the Query tab, review the query, variables, and metadata.
    • You can also test the job using synthetic data and view the test results to validate the logic before submitting your review.
  3. Click Accept Job or Reject Job.
    • If rejecting, you must enter a comment.
    • Comments are visible to all workspace members.
  4. Click Submit. You cannot change your review after submission.

Note

If the job has input variables, you will have to check both Reviewed? boxes at the bottom of the Input Variables tab and the query editor before you can approve the job.

Step 3. Run your job

Once all workspace members approve a job, its review status changes to Accepted, and all members are notified. At this point, the job is locked—no further changes can be made.

Only the job creator can run the approved job:

  1. Go to the Query tab and click Run Job.
  2. Note that the job's status changes to Queued.

The job remains Queued if another job is currently running in the workspace. Once execution begins, the status updates to Running.

Note

If your job includes input variables, you’ll be prompted to enter values before execution. For details, see Define Input Variable values.

Step 4. Retrieve the results

Once your job runs successfully, the results are encrypted and saved to your organization's storage. When the status changes to Succeeded, you can view or export the results directly from the job page. If the job fails or is canceled, no results are produced—but you can check the logs to understand what went wrong.

View logs

Only workspace members from the job creator’s organization have access to the job logs.

  1. On the Query tab, select Logs to view job execution details.
  2. If the job failed, error information appears at the top of the log.
  3. Click Export Logs to download the logs as a text file.

Export results

Only workspace members from the job creator’s organization can view the job results:

  1. On the Query tab, select Results to preview the first 100 rows of the job results.
  2. Click  Export Results to download the results as a CSV file.

Rerun a Job

Once your job is approved, you can rerun it anytime—no need to resubmit for review. Reruns are useful for recurring workflows, testing across data snapshots, or refining results with new inputs.

You can rerun a job as-is or update input variables and datasets before execution. Each rerun is tracked in the job’s Run History, allowing you to compare outputs across executions. For details on managing reruns, see Job reruns.