The increasing complexity of software systems has intensified the need for strategies that improve their internal structure without altering functionality. In this context, remodularization emerges as a key practice to reorganize software components in order to enhance maintainability, scalability, and comprehensibility. This study proposes a multi-objective optimization approach based on genetic algorithms to improve the modular quality of complex systems. The method focuses on minimizing coupling and maximizing cohesion as primary quality criteria. To achieve this, both static and dynamic analysis techniques are employed to reconstruct the existing architecture and evaluate alternative modular configurations. The results show significant improvements in software structure, including reduced coupling and increased cohesion, as well as a better distribution of responsibilities among modules. Furthermore, findings indicate that there is no single optimal solution, but rather multiple trade-off alternatives between the considered objectives. Expert validation supports the feasibility and usefulness of the generated recommendations. Overall, the study demonstrates that integrating evolutionary optimization techniques with software analysis provides an effective tool to support maintenance and evolution processes in complex systems, particularly in research software contexts.