I used to train AI to help researchers find more relevant papers at http://iris.ai This was nothing more than just classifying. Would this kind of opportunity be available for data remotaskers at Scale. Best regards for groundbreaking work