Flooding is a complicated and deadly natural phenomenon that causes significant harm to infrastructure, property, and human life. Through the use of machine learning and satellite imagery, this study offers a novel method for estimating flood risk. A model was created to detect possible water channels in satellite photos using the FastAI package and a public accessible dataset from Kaggle. This makes it possible to predict high-risk locations for flooding more precisely. In order to facilitate user access to the tool, Streamlit, an online application, was developed. This allows users to upload satellite photos and obtain predictions on the percentage of water covered. Authorities may more effectively prioritize disaster response and distribute resources to the places that need them most thanks to the application's vital information severely damaged by floods. This approach shows how machine learning and remote sensing can be combined to improve the way flood risk is assessed. With the potential to save lives and lessen the overall impact of such disasters, this strategy offers a workable way to improve flood preparedness and response by combining satellite data with sophisticated image processing techniques.