In this work we show preliminary results of deep multi-task learning in the area of computational pathology. We combine 11 tasks ranging from patch-wise oral cancer classification, one of the most prevalent cancers in the developing world, to multi-tissue nuclei instance segmentation and classification.