In most cases, brain tumors have a longer lifespan than normal brain cells because they are defined by an abnormal multiplication of brain cells. Timely detection and accurate diagnosis of these cells can avert brain tumors from becoming life-threatening. The field of computer-aided diagnostic techniques has seen extensive research in the exploration and classification of brain tumors. Recently, the CNN, a deep learning technique for image classification, has gained significant traction for its effectiveness in the medical domain. This study highlights the application of deep learning as well as machine learning techniques for intracranial tumor classification, utilizing the publicly accessible figshare dataset. We employ a custom-designed CNN architecture for feature extraction, coupled with an ensemble of several machine learning models for the classification of brain tumors based on the extracted features. Our model demonstrates superior execution with an overall accuracy of 95.65%, surpassing numerous contemporary deep learning models.