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

# Data Connectors

To train models, you first need to connect a data source. This enables Kumo to train and retrain models effortlessly, making predictions as simple as refreshing your data.

## Set up Connector

1. Go to **Connectors** from the side menu and click **Add Connector** to create a new connection.
2. Select your data source and provide the required configuration details.

![](/sdgm/_files/nvidia-sdgm.docs.buildwithfern.com/b06ba486c9b9e3b0f6ee2a1955da8803f7771cd1fc6f257cb26f1b9d840670ff/img/connection-options.png)

### Supported Data Sources

Kumo currently supports the following data source types:

* [AWS S3](/aws-s3): Reads Parquet (recommended) and CSV files; requires IAM role-based access.
* [Google Cloud BigQuery](/google-cloud-bigquery): Reads tables; requires security credentials.
* [Snowflake](/snowflake-connector): Reads tables and views; requires security credentials.
* [Databricks](/databricks-connector): Reads tables; requires security credentials.

- [AWS Glue Catalog:](/aws-glue) Reads Parquet files on S3 via a Glue Catalog; requires IAM role-based access.

For local file uploads, select [Local Upload](/local-data-upload) to create a connector, then click **Connect Table** and upload the files.

## Managing Connectors

To edit or delete connectors, select the Connector → click the **︙options button** and choose **Edit** or **Delete**.