Machine learning models and deep learning techniques have been proven to be quite helpful to clinical professionals accurately diagnose many medical conditions. Inspired by the recent use of machine learning models, we develop a deep learning model that detects intracranial hemorrhages in CT brain scans. The model classifies the scans into two classes, hemorrhagic and normal. We evaluate this model using multiple public datasets. Our model 1) increases the speed of diagnosis, removing the significant dependency on highly trained professionals 2) reduces the cost for diagnosis, 3) increases accessibility in rural areas, where highly trained specialists may not be found, and 4) reduces the strain on overworked radiologists that have to go through thousands of CT scans a day.