The objective of this project was to develop a neural network model for porous media fluid flow using training data gathered from readily available direct numeric simulation software post processed by dynamic mode decomposition. The driving reason for the development of a neural network model for fluid flow simulation applications is the computational cost associated with dense meshes required for accurate simulations. With a neural network model of the dynamics associated with the before mentioned porous marital fluid flow, much faster simulations can be run with greatly decreased computational cost.