Abstract As emissions regulations for greenhouse gas emissions become more strict, it is important to increase the efficiency of engines by improving on the design and operation. Current optimization methods involve performing large numbers of experimental investigations on physical engines or making use of detailed Computational Fluid Dynamics modeling efforts to provide visual and statistical insights on in-cylinder behavior. The latter still requires experimental data for model validation. Both of these methods share a common set of problems, that of being monetarily expensive and time consuming. Previous work has proposed an alternative method for engine optimization using machine learning (ML) models and experimental validation data to predict scalar values representing different parameters. With such models developed, one can then quickly iterate on operating conditions to find the point that maximizes an application-dependent reward function. While these ML methods provide information on individual performance parameters, they lack key information of in-cylinder indicators such as cylinder pressure traces and heat release curves that are traditionally used for performance analysis. This work details the process of implement- ing a Multilayer Perceptron (MLP) model capable of accurately predicting crank-angle resolved high-speed in-cylinder pressure using equivalence ratio, fuel injection pressure and injection timing as input features. It was demonstrated that the model was able to approximate engine behavior with mean squared error lower than 0.05 on a 1-55 range in the test set. This approach shows potential for greatly accelerating the optimization process in engine applications.