I have a dataset sitting in Azure CosmosDB for MongoDB. It is about 3.5 million records. I have a Python script that looks for duplicates in the dataset. It is basically an aggregation pipeline, here is the snippet of the code: