ABSTRACT : We presented our solution for the SIIM-ISIC melanoma classification challenge. This is a multi-class multi-modal classification model using images and metadata and, we tested both binary and multi-class image-only models and a binary multi-modal model. The keys to success for our solution were the selection of the target variable, using the available metadata, and the data augmentation strategy. Achieving AUC values of 0.95 and F1 of 0.71 for the validation data.