How to build an inferencing engine