Alzheimer’s disease (AD) is a disorder that leads to deterioration of memory and loss of cognitive functions. It severely impacts a person’s ability to perform day to day tasks.Traditionally diagnosis is done through subjective or invasive biomarkers, which is neither comfortable nor easily accessible. This paper ponders upon the possibility of using 3D Convolutional Networks (3D-CNNs) and Conventional Autoencoders (CAEs) for early AD detections from MRI scans. The framework showcases improved classification performance by focusing on subtle anatomical changes such as hippocampal atrophy. The advanced pre-processing techniques, feature extraction, and dimensionality reduction are integrated to overcome shortcomings like imbalanced datasets and computational demands. The findings help to navigate for more scalable, efficient and accurate diagnostic solutions to aid clinical work and public health efforts.