> For clean Markdown of any page, append .md to the page URL.
> For a complete documentation index, see https://docs.umnai.com/llms.txt.
> For AI client integration (Claude Code, Cursor, etc.), connect to the MCP server at https://docs.umnai.com/_mcp/server.

# Onboarding jobs

An onboarding job processes an available datasource into a dataset. Prepare an [onboarding configuration](/ui-guides/data/onboarding-configuration) and a processing compute configuration before submitting the job.

## Before you begin

Confirm that the datasource is available, the onboarding configuration matches the prediction task and a [processing compute configuration](/ui-guides/setup/compute-configurations) exists. Choose **Initial** for a new dataset or **Incremental Update** for an existing dataset. The job type also determines which configurations are offered.

## Run an initial onboarding job

1. Open **Datasets > Onboarding Jobs** and select **Create**.

2. Complete the following fields, then select **Create**.

| **Field**                        | **What to enter or select**                                  |
| -------------------------------- | ------------------------------------------------------------ |
| Name                             | A unique name for the onboarding job.                        |
| Job type                         | Initial for a new dataset.                                   |
| Dataset name                     | A unique name for the output dataset.                        |
| Datasource                       | The available datasource to process.                         |
| Dataset onboarding configuration | The saved configuration prepared above.                      |
| Compute configuration            | The processing compute configuration for the onboarding job. |
| Maximum runtime minutes          | The job time limit; the displayed starting value is 60.      |

### Review the selected inputs

Complete the datasource and configuration selections before submitting the job.

![Initial onboarding job form before completion](/_fern-files/_img/606e6f0b04d1a4bdc1c1cd34f85928c7952f422d60661bcd26f9bd2e79acd5cf.webp)

## Check onboarding completion

1. Open the job and inspect **Status**, **Progress** and **Stage**. Check the source datasource, target dataset and configuration references. The completed example shows **COMPLETED**, **100%** and **Persistence**. If processing fails, inspect the error and verify the source and configuration before running the job again.

2. When processing succeeds, open **Datasets** and verify that the output dataset is **AVAILABLE**.

![Existing completed onboarding job](/_fern-files/_img/02183d92387ebe163d4dcdc9140337a69c0cb61d1dff1ddac7e8de5c1147608e.webp)

## Update an existing dataset

1. Open **Onboarding Jobs > Create** and enter a job name.
2. Select **Incremental Update** and choose the existing dataset. Optionally supply an **Increment label** to identify the update.
3. Select the datasource containing the additional data and an incremental onboarding configuration.
4. Select processing compute and set a positive whole-number runtime in minutes.
5. Review the resource references, select **Create** and monitor the resulting job.
6. Inspect the output dataset and its increment and lineage information when processing completes.

Changing job type resets type-specific selections. If a configuration disappears, check that its type matches the selected job. An onboarding update changes the dataset context; train a new model when you need a model built from the updated data.

## Next steps

Continue to [Inspect datasets](/ui-guides/data/datasets).