Consuming a Registered Model inside Data Pipeline Module

Prerequisites:

Before you can use a DSL model in a pipeline, ensure that:

1. A model has been trained and saved from a Data Science Lab notebook.

2. The saved model has been registered in the Models section (so it is available to other modules).

3. You have permission to create or edit pipelines in the Data Pipeline / Designer module.

Create or Open a Pipeline

• In the left navigation, go to Designer (or Data Engineering → Pipelines, depending on your UI).

• Create a new pipeline or open an existing one.

• Give the pipeline a meaningful name, for example, Sample Pipeline (as shown in the screenshot).

You should now see an empty pipeline canvas.

Add the DSLab Runner Component

• On the right-hand side, open the Components panel (if it is not already open).

• Expand the Machine Learning section.

• Locate the DSLab Runner component.

• Drag and drop the DSLab Runner onto the pipeline canvas.

  

This component is responsible for invoking a DSL model from within the pipeline.

Open the DSLab Runner Configuration

• Click the DSLab Runner node you just placed on the canvas.

• A configuration panel appears at the bottom of the screen with two tabs:

  • Basic Information

  • Meta Information

• Select the Meta Information tab (highlighted in the screenshot).

All model-specific settings are configured here.

Set Execution Type, Project, and Model

In the Meta Information tab, fill in the following fields:

• Execution Type

  • From the dropdown, select Model Runner.

  • This tells the component that it should execute a registered DSL model (rather than a script or notebook).

• Project Name

  • Choose the Data Science Lab project where the model was created and registered.

  • Example: Sample Project.

• Model Name

  • From the dropdown, select the registered model you want to consume in this pipeline.

  • Example: Sample Saved Model.dill.

  • The list shows all models registered and available for that project.

    

Once these values are set, the DSLab Runner is linked to the chosen registered model.

Connect the DSLab Runner to Other Components

To use the model as part of a working pipeline:

• Add Reader components (e.g., database, file, or stream readers) to fetch input data.

• Connect the Reader(s) to the DSLab Runner node so that the model receives the required features as input.

• Add appropriate Writer components (e.g., database writer, file writer, message queue) and connect them to the output of the DSLab Runner to store or forward prediction results.

This creates a flow such as:

Reader → DSLab Runner (Model Runner) → Writer

Save and Validate the Pipeline

• Click Save (toolbar icon) to persist the pipeline definition.

• Optionally run a validation (if available) to ensure all mandatory fields, connections, and schema mappings are correct.

Execute the Pipeline

• Use the Run / Play button in the pipeline toolbar to start execution.

• Monitor the job status from the pipeline console or job monitor:

  • Successful execution confirms that the registered DSL model was invoked correctly.

  • Any errors (e.g., missing features, schema mismatch) will be shown in the logs.

When the pipeline runs successfully, the registered DSL model is now fully operational inside the Data Pipeline, performing inference on incoming data and delivering results to the configured targets.

Summary

To consume a registered DSL model in the Data Pipeline module, you:

• Register the model in Data Science Lab → Models.

• Create/open a pipeline and add a DSLab Runner component.

• Set Execution Type = Model Runner, select the Project Name, and choose the Model Name.

• Connect Readers → DSLab Runner → Writers.

• Save and execute the pipeline.

This allows you to move seamlessly from notebook experimentation to production-grade, pipeline-driven ML inference inside the BDB platform.