Sustainable groundwater management is essential to mitigate the adverse impacts of climate change and ensure freshwater availability. This study introduces a predictive and recommendation system that forecasts groundwater levels using Long Short-Term Memory (LSTM) networks and classifies seasonal patterns using K-Means clustering. The data-driven approach uses information collected from the “Pantir” monitoring initiative at Universitas Indonesia. The LSTM model, optimized with Grid Search and early stopping, achieved an R2 of 0.9211, MAE of 6.1028, and RMSE of 9.3159. The clustering component identifies Indonesia's wet and dry seasons, enhancing the contextual accuracy of the prediction results. This study offers a novel integration of LSTM forecasting and K-Means classification within a user-friendly dashboard. The system is intended to support both policymakers and the general public by providing accessible insights that encourage sustainable groundwater usage and informed water resource planning. © 2025 IEEE.