Skin cancer remains one of the most widespread cancers globally, and detecting it early plays a vital role in ensuring effective treatment. However, traditional diagnosis methods depend heavily on the expertise of dermatologists, which can make the process slow and costly. This project introduces an automated approach to skin cancer detection using a combination of deep learning and machine learning techniques, aimed at supporting early and efficient diagnosis. To improve accuracy and reliability, several preprocessing steps were applied, including image augmentation, normalization, and class balancing. The model was further enhanced using transfer learning with pre-trained ImageNet weights, allowing it to perform well even with limited data.