Can hopsworks feature store be customized as stand alone to fit other existing ML platform enterprise level?
Yes, sure. It is widely used as a standalone feature store from platforms such as Databricks, Sagemaker, Azure ML Studio. It can be used from any Python or Spark environment - you compute your features in Python, Spark, Flink and write them to the feature store. You can use APIs to retrieve or save training data, and for serving there are APIs to retrieve cached features as feature vectors. You can also use an external data warehouse as a source for your offline features, by defining tables, for example in Snowflake, as “on-demand feature groups”.
See more here: