Built a multivariate LSTM model in Python to forecast air pollution using time series data. Preprocessed data with lag features, normalized inputs, and reshaped for supervised learning. Trained on multiple timesteps, evaluated with RMSE & MAE, and delivered a scalable forecasting pipeline.