This study introduces a deep learning based artificial intelligence methodology to investigate and optimize a jet impingement cooling (JIC) arrangement. JIC systems are widely used to enhance heat flux in high-temperature applications by directing jets onto the surface. However, the effectiveness of this method can be limited due to the deflection of jets caused by crossflow effects, and jet-to-jet interactions. The present research uses artificial intelligence-based methods to predict and optimize an impinging jet configuration in the presence of cross flow and jet-to-jet interaction. A linear, equidistant array of extended jets is investigated that impacts on a wavy surface, whereas a cross flow is imposed through the space between the jets and the surface. The effects of the varying jet lengths and the corrugation depth of the target wavy wall are analyzed. The input data for the artificial intelligence model are obtained from computational fluid dynamics-based numerical flow field simulations. The simulations are performed for three Reynolds numbers (Re) (15000, 25000, and 35000), various jet lengths (L) from 0 to 25 mm, and corrugation depths (H) ranging from 0 to 10 mm. To optimize the thermal performance, artificial intelligence-based optimization methods, such as Nelder-Mead, Powell, SLSQP, and TNC, are employed. The Nusselt numbers (Nu) of the optimal configurations suggested by these algorithms are compared with those found by computational fluid dynamics for the suggested configurations. The results demonstrated that the model of Powell and Nelder–Mead algorithms produced the highest average Nu. In addition, the configurations of the TNC algorithm presented the most uniform local heat transfer distribution and stable thermal performance. In conclusion, optimization with AI significantly improves the design of JIC systems on complex surfaces.