Question

Your ML platform team maintains a legacy data-validation tool that is already packaged as a Docker image with a well-defined command-line interface (it reads an input GCS path flag and writes a validation report to an output GCS path flag). You want to reuse this exact container inside a new Vertex AI Pipeline authored with the KFP SDK v2, without rewriting the logic in Python or rebuilding the image. What is the most appropriate way to incorporate this tool as a pipeline step?

Accepted Answer

Define the step with a container component (e.g., @dsl.container_component) that specifies the existing image and maps pipeline inputs/outputs to the tool's command-line arguments.

Answer

Rewrite the validation logic as a lightweight Python function component decorated with @dsl.component so KFP can auto-build a container at compile time.

Answer

Deploy the container as a Vertex AI online endpoint and call it from a Python component using the prediction REST API.

Answer

Load the Docker image into Cloud Run and trigger it from the pipeline with an Eventarc GCS event so validation runs out-of-band.