This work introduces a distributed random parameter optimization model for short-term multivariate time series forecasting. It combines deep neural networks for forecasting with distributed hyperparameter tuning, enhancing resource utilization and computational speed.