<p>Fused Deposition Modeling is a widely used additive manufacturing technology, but the mechanical properties and dimensional accuracy of its products are highly dependent on the selected process parameters. However, existing research typically employs single optimization algorithms in isolation, leaving practitioners with no empirical basis for selecting between fundamentally different approaches when facing real-world constraints of time, computational resources, and the need for either comprehensive exploration or rapid convergence. This study employs a multi-objective optimization approach, utilizing both a Non-dominated Sorting Genetic Algorithm and Bayesian Optimization, to determine the optimal set of FDM parameters for printing Polylactic Acid specimens. The objectives are to maximize Ultimate Compression Strength (UCS) and minimize the compression average percentage deviation (dimensional accuracy). Twenty-five specimens were fabricated and tested according to ASTM D695 standards, varying six parameters: infill density, pattern, and overlap, layer thickness, shell thickness, and top/bottom layer number. The results demonstrate a fundamental trade-off between strength and accuracy. The optimization successfully identified Pareto-optimal solutions, with the best solutions achieving a UCS of 60.00&#xa0;MPa and a minimum deviation of 0.400%. Key findings confirm that infill density is the most significant parameter for strength (78.856%), while layer thickness is paramount for accuracy (37.013%). While NSGA-II offered better solution diversity, Bayesian Optimization proved to be 6.3 times faster, making it more suitable for practical, time-sensitive applications. This research provides a clear framework for selecting FDM parameters based on application-specific requirements for strength, accuracy, or a balanced compromise. The novelty lies in the first direct comparison of these algorithms for FDM optimization, providing evidence-based guidance for algorithm selection based on project priorities.</p>