Computer-Aided Design (CAD) is the most prevalent approach to modeling objects for manufacturing: from coffee mugs to sports cars. This process is complex and requires years of training to master. One of the elements of CAD that is especially difficult to get right is sketching. Sketches are highly structured 2D drawings that lie at the heart of every 3D construction. In this work, we set out to tackle the problem of automatic generation of such drawings and through that enable development of intelligent tools helping humans create better designs with less effort. Our method is a combination of a general-purpose language modeling technique alongside an off-the-shelf data serialization protocol. We show that while being simple our approach is flexible enough to accommodate the complexity of the domain and demonstrates good performance both in case of unconditional synthesis as well as in the image-to-sketch scenario.