The World Health Organization (WHO) reports a notable increase in malignant skin cancer cases during the last ten years. Over 150 thousand cases of skin cancer have been detected around the world. It is necessary that skin cancer is detected and diagnosed in its initial stage4. Skin cancer is mostly caused by smoking, UV radiation exposure, DNA alterations, and bad lifestyle choices. Clinical screening, dermoscopic analysis, histological investigation, and biopsy are commonly used in the diagnosis process. However, even experienced dermatologists find it difficult to spot problems early. The usefulness of deep learning and transfer learning in the categorization and diagnosis of skin cancer has been highlighted by recent studies. In the present study, we use deep learning techniques to present a successful approach for categorizing skin cancer. Specifically, using the HAM1000 skin lesion dataset, researchers modified the Mobile Net CNN work that had already been trained. Notable results were obtained in terms of average-weighted precision, recall, and category accuracy using our transfer learning strategy utilizing InceptionV3: 93%, 92%, and 94.7%, respectively. This lightweight and fast model offers reliable support for dermatologists in early skin cancer prognosis.