Previously, this was using a pre-defined pipeline that had to be loaded into the kubernetes context before this ran. Now, the pipeline is inlined with resolvers for the various tasks used from the catalog.