In this paper, at the same time, two separate objectives and two safety and operational constraints are chosen to optimize fuel reloading pattern of a Material Testing Reactor (MTR), independently and coherently. This is one of the most difficult type of engineering problems as a constrained, non-continuous, combinatorial, and fully multi-objective optimization problem. Decision space is a non-continuous multimodal space restricted by both of the combinatorial and safety constraints. A smart software application and a robust hybrid algorithm have been developed to get Pareto optimal set with respect to both of the economy of irradiating utilizations and nuclear safety based on the heuristic soft computing. The hybrid algorithm is composed of a fast and elitist Multi-Objective Genetic Algorithm (MOGA) and a fast fitness function evaluating system based on the semi-deep learning cascade feed forward Artificial Neural Networks (ANNs). The smart software is used to produce database automatically required for the ANN training and test data. It can be also used to revise data accurately, impose further irradiating benefits or Operating Limits and Conditions (OLCs), and to advise the reactor supervisor on the most desire pattern based on the smart searches and filtering. The results are highly promising. For more details, optimization results dominate conventional operating core parameters, significantly. Also chosen OLCs are protected. Furthermore, this is very good practice to reach a fully developed practical application of the complex soft computing for the nuclear fuel management problems.