What Is Wallaroo?

Wallaroo is a machine learning operations (MLOps) platform designed to help organizations operationalize their ML models at scale. Businesses use Wallaroo to deploy, manage, and monitor models in production—whether those environments are on-premises, in their own cloud infrastructure, or in a hybrid setup. Because each customer runs a dedicated Wallaroo instance, configuration details such as endpoints, authentication, and infrastructure differ across implementations. This starter kit provides a lightweight, branded foundation that customers and services teams can extend to support their unique model workflows, inference requirements, and data pipelines.

---

Product Type: MLOps

Integration Type: Starter Kit

Event Source Type: Web, Mobile Browser, & Mobile App

Event Scope: Full-Funnel Events

---

Capabilities

• Sends MetaRouter events to a customer-owned Wallaroo inference endpoint for model scoring.
• Uses a generic inputs structure so customers can adapt features per model and use case.
• Works with many event types (identify, page, and common track events) via a shared payload pattern.
• Provides a lightweight starting point that customers can customize for their own ML pipelines.

---

Considerations

• Each customer runs a separate Wallaroo instance; endpoint URLs, realms, and credentials are unique per environment.
• The starter kit sends data to the configured inference endpoint; all routing, storage, and model lifecycle take place in Wallaroo.
• Required parameters and feature layout depend on each customer’s model design.
  • click_data and site_id in this starter kit are placeholder feature fields. These values are not required by Wallaroo itself and can be renamed, expanded, or removed entirely depending on your ML pipeline and use case/
  • If you do choose to use click_data, note that click_data must contain six values in the expected format; otherwise, the Wallaroo endpoint may reject the request or return an error.

---

Limitations

• No detailed feature schema is provided; customization is required for each customer model.
• This integration only transports data to an inference endpoint; it does not manage or deploy ML models.
• Predictions are not automatically persisted or fed back into MetaRouter.
• Full end-to-end validation must be performed by your ML project team.

---

Identity Sync

• Sync Available: No
• Required for Starter Kit: No

---

Starter Kit Setup Guide

1. Gather Credentials

Ask your Wallaroo or internal platform team for:

• External inference endpoint URL
• Wallaroo instance base URL
• Confidential client ID
• Confidential client secret
• Access Token

2. Add a Wallaroo Integration

3. Configure Event Mapping

MetaRouter provides generic mappings for Wallaroo. Add or adjust mappings as needed for your model.

4. Deploy to Pipeline

• In Pipelines, add the Wallaroo integration.
• Select the correct revision.
• Click Add Integration to deploy.

---

Event Mappings

*Please customize all mappings as needed for your model.

Global

Event Specific

All supported events use the same global inputs mapping. Please customize as needed.

---

Integration Validation

To validate the integration, ensure that events sent from MetaRouter are successfully delivered to your Wallaroo inference endpoint—for example, by confirming that the endpoint returns a successful response such as HTTP 200. After delivery is verified, you should confirm within your Wallaroo environment that the incoming data is being ingested by the appropriate pipeline and processed according to your model configuration. Because model logic, feature requirements, and inference outputs vary by customer implementation, MetaRouter cannot validate model execution or prediction results. All model-level validation must be performed within your Wallaroo instance.

---

Additional Wallaroo Documentation

• Wallaroo MLOps Connection Guide

---