This is a bangla auto-complete engine (basically a next predictor built from scrach . no nltk no scikit learn). Based on n-gram model. Gives a bit of akward output as the corpus that is used to build this, mainly conatains news from bengali news papers. a django app is built on top of it to make it more user friendly. Can be used as an API too as it has some RESTful property. I have done done some pickling(writing object to file) just to fast things up.