This document describes our approach to building an Offensive LanguageClassifier. More specifically, the coursework required us to build threeclassifiers with slightly different goals: - Offensive language identification: would classify a tweet as offensive ornot. - Automatic categorization of offense types: would recognize if the target ofthe offense was an individual or not. - Offense target identification: would identify the target of the offensebetween an individual, group or other. In this report, we will discuss thedifferent architectures, algorithms and pre-processing strategies we tried,together with a detailed description of the designs of our final classifiersand the reasons we choose them over others.