The Tagotieb gossans near Derudeb in the Red Sea Hills of northeastern Sudan occur within the Haya Terrane of the Arabian-Nubian Shield, in a geological setting prospective for volcanogenic massive sulphide (VMS) mineralisation. Two principal gossan bodies are hosted by foliated acidic metavolcanic rocks, together with associated altered granite and gabbro. Field mapping, petrographic analysis, and ore microscopy indicate that goethite, with subordinate hematite and other alteration minerals dominate the gossans, and that they preserve relict sulphides and native gold. Brecciated, botryoidal, colloform, and boxwork textures suggest the development of a mature, largely indigenous gossan formed above a polymetallic sulphide system. Landsat-9 and Sentinel-2 surface reflectance data were analysed using mineral-sensitive band ratios, spectral indices, false-colour composites, and principal component analysis to enhance iron oxides, clay minerals, barite, and silica-rich zones. Supervised Support Vector Machine (SVM), Mahalanobis Distance (MD), and Artificial Neural Network (ANN) classifiers, trained using field-validated samples, delineated gossans and host lithologies with overall accuracies of 78%, 74%, and 69%, respectively. The spatial distribution of alteration defines a NE-SW-trending corridor intersected by NW-SE-oriented structures controlled by regional tectonics. This configuration supports a three-stage genetic model involving sulphide oxidation, Fe-oxyhydroxide evolution, and subsequent mechanical reworking. These results demonstrate that multispectral indices combined with supervised machine-learning approaches provide a practical and cost-effective framework for targeting VMS-related gossans in the Red Sea Hills and comparable Neoproterozoic terranes.