Bengali.AI is a community working to solve the absence of open-sourced datasets for Bengali Computer Vision/Natural Language Processing research. Bengali.AI has launched a pilot competition through which we are also making our first dataset public, which has more than 85,000+ Bengali Handwritten Digits. It’s a month-long online machine learning competition where the participants have to train their computer/model on our open-sourced database ‘NumtaDB’ to be able to detect handwritten Bengali digits (০,১,২,৩,...) and test the performance on a hidden test set through Kaggle’s online platform. It is the first of a series of competitions called “Bengali.AI Computer Vision Challenge” through which we wish to open-source more datasets in the future.