This project focuses on predicting oil, gas, and water production rates using a Gated Recurrent Unit (GRU) neural network. The historical dataset obtained from DSEATS fields contains 16 features and 6,925 data points. Exploratory data analysis revealed missing values, single-value columns, and negative water production values, all of which were handled through feature engineering and data preprocessing. The preprocessed dataset was split into training and test sets, then scaled and converted into time series format with 10 time steps. A GRU model was selected for prediction after tuning its hyperparameters using GridSearchCV. The model's performance was validated, achieving a Mean Squared Error (MSE) of 0.002, Root Mean Squared Error (RMSE) of 0.048, and Mean Absolute Error (MAE) of 0.023. The results were visualized through a graphical comparison of actual and predicted production rates, showcasing the model's predictive accuracy.