This thesis focuses on classifying Age Related Macular Degeneration (AMD) into wet (Exudative) and dry (Non Exudative) by analyzing 3D OCT scans with Machine learning. Due to the sparse size of the dataset, which contains fewer than 100 samples, a two part design was proposed and applied in this thesis. The two part solution consists of an autoencoder for feature extraction and dimension reduction, and a classifier for the final prediction. This approach aids in overcoming the limited size of the dataset while capturing important features from the volumetric OCT scans. This projects shows a complete pipeline for medical imaging processing using machine learning, from data cleaning and preprocessing, to model design, to training and evaluation.