Process control based on physics-based modeling requires detailed real-time reactor simulations, which are currently not realistic. For such process control models to be feasible, information from reactor simulations must therefore be represented in a compact model. In this paper, we have developed a neural network based model for chemical vapor deposition. Detailed reactor simulations are used to train the neural network and the network predictions are then validated by additional simulations. We show that the current model is capable of accurately representing the process parameter space, thereby enabling use of the trained network for process control and design.