In one embodiment, a device, in communication with a plurality of data lake sites, receives a federated data lake query. The device determines a plurality of data lake operator sets that each correspond to one of the plurality of data lake sites, wherein each of the plurality of data lake operator sets is used to establish a respective data pipeline for the federated data lake query. The device selects a particular data lake site of the plurality of data lake sites as a destination for data pipelines that are established for the federated data lake query. The device sends the plurality of data lake operator sets that each correspond to one of the plurality of data lake sites to cause the plurality of data lake sites to send query results to the particular data lake site using the data pipelines, wherein the particular data lake site stitches the query results.