Direct Dual Fuel Stratification (DDFS) is a novel LTC strategy among other strategies which uses two direct injectors in the combustion chamber, similar to Reactivity-Controlled Compression Ignition (RCCI), but resulting in more authority over the combustion process and the rate of heat release. DDFS has comparable thermal efficiency to RCCI and HCCI, as well as extra-low NOx and soot emissions, and it also is able to meet the EURO6 emission mandate without using aftertreatment under optimized conditions. Thus, it is crucial to optimize the injection strategy of both injectors in a DDFS engine. Artificial Neural Networks (ANNs) are used to develop a model for predicting engine performances and pollution. A multi-objective optimization analysis was performed to minimize NOX, soot and fuel consumption simultaneously using the non-dominated sorting genetic algorithm (NSGA-II) for the injection parameters of the gasoline and diesel direct injectors. The optimal solutions met the EURO6 mandate for NOX and soot, offered lower fuel consumption up to 8 g/kW-h, and also had about 2% higher thermal efficiency than the base case. Thermodynamic evaluation based on the first and second laws were performed for seven selected candidates on the Pareto Front and compared with the base case.