Skin cancer can manifest in a variety of ways and requires specialized knowledge to accurately classify, diagnosing it can be extremely difficult for medical professionals. Effective diagnosis is frequently hampered by the complicated nature of skin lesions and the requirement for specialized knowledge. Using convolutional neural networks, the proposed work offers a dependable and efficient method for classifying skin cancer. Deep learning is used by the CNN model to automatically recognize intricate patterns and characteristics. The approach enables the precise and automated categorization of skin lesions into several groups. The "Skin Cancer MNIST HAM10000" dataset, which consists of 10,000 images of various skin lesions, is the specific dataset that the proposed model is trained on. This large and diverse dataset contains several forms of skin cancer images. To improve performance even more and enable the model to generalize effectively over a variety of skin textures, lighting conditions, and lesion sizes, data augmentation techniques are applied. Apart from its ability to detect skin cancer more quickly, the proposed CNN-LSTM based approach also shows remarkable accuracy, with an overall accuracy rate around 97%. Since the number of occurrences of skin cancer is rising, the automatic classification system is a useful tool for medical professionals. Early detection aids in the fight against this severe sickness, which ultimately leads to progress.