Concurrency at every stage of the data processing pipeline...

Process any number of files, of any length, for any number of filters

Multitask across all input files, broadcasting every row concurrently

Evaluate each row against any number of filters per input file

Dynamic data type inference and concurrent data parsing

Robust error handling of potentially unclean input data

Multiple date formats supported via user defined format key library

Generate user defined named tuples classes for each row processed

Multi-file output supports custom output directory and file names