Study region: Siwa Oasis in Egypt faces critical water management challenges due to its arid climate, complete dependence on groundwater, and rapid aquifer salinization. The Oasis relies entirely on two main aquifers, the shallow Tertiary Carbonate Aquifer (TCA) and the deep Nubian Sandstone Aquifer System (NSAS) with no significant recharge due to minimal rainfall and extreme evaporation rates. Study focus: This study integrates Self-Organizing Maps (SOM), k-means clustering, and hydrogeological methods to characterize the NSAS and estimate critical hydraulic parameters. Geophysical well logs were analyzed using unsupervised machine learning techniques to delineate aquifer layers and construct a novel 3D facies model. The Kozeny-Carman equation was applied for hydraulic conductivity estimation and validated against pumping test data. SOM clustering identified distinct lithological units and depositional environments, while 3D modeling mapped subsurface heterogeneity across porosity, shale volume, and hydraulic conductivity distributions. New hydrological insights for the region: Results reveal hydraulic conductivity ranging from 1.2 to 6.6 m/d, effective porosity between 24 % and 40 %, and shale volume from 0 % to 60 % across the NSAS. The central and eastern parts of Siwa Oasis emerge as optimal zones for sustainable groundwater extraction. High-porosity channel sands were identified as primary aquifer units, while finer-grained floodplain deposits act as flow barriers. This represents the first systematic application of machine learning for aquifer characterization in Siwa Oasis, providing essential data for sustainable groundwater management and future drilling.