Flooding poses significant risks to communities, ecosystems, and infrastructure, necessitating precise forecasting systems, especially in high-risk regions. This project focuses on flood forecasting for the Nellore city in Andhra Pradesh, which experiences sudden flash flooding due to intense monsoon rains. Traditional hydrological models for flood prediction often require extensive data and calibration, presenting challenges in regions with limited resources. Advances in machine learning (ML) and deep learning (DL) offer a promising alternative, capable of processing large datasets and identifying complex, nonlinear relationships within hydrological data. This project leverages hybrid deep learning frameworks to enhance flood prediction and forecasting accuracy for the Nellore city located in the Pennar River Basin. We utilize three primary approaches: a standalone LSTM model, a standalone CNN model and a hybrid LSTM–CNN model. The chosen model variables include hydrometeorological, geological and remote sensing data derived from HydroSHEDS, WRIS and IMD, capturing critical features such as precipitation, temperature, wind speed, wind direction, elevation, slope, aspect, land use and land cover and historical discharge data. Data preprocessing techniques, including lagged features, cyclic transformations and hyperparameter tuning, are applied to improve model performance. This model's adaptability to different type of floods and consideration of rare occurrences, positions it as a valuable tool for local authorities, enabling proactive responses to imminent flood threats and supporting disaster risk reduction efforts in the Nellore city of Andhra Pradesh. The project showcases AI & ML’s potential to enhance flood forecasting accuracy, providing critical insights for regions vulnerable to climate-driven hydrological changes.