Optimization of metabolic pathways consisting of large number of genes is challenging. Multivariate modular methods (MMMs) are currently available solutions, in which reduced regulatory complexities are achieved by grouping multiple genes into modules. However, these methods work well for balancing the inter-modules but not intra-modules. In addition, application of MMMs to the 15-step heterologous route of astaxanthin biosynthesis has met with limited success. Here, we expand the solution space of MMMs and develop a multidimensional heuristic process (MHP). MHP can simultaneously balance different modules by varying promoter strength and coordinating intra-module activities by using ribosome binding sites (RBSs) and enzyme variants. Consequently, MHP increases enantiopure 3S,3'S-astaxanthin production to 184 mg l-1 day-1 or 320 mg l-1. Similarly, MHP improves the yields of nerolidol and linalool. MHP may be useful for optimizing other complex biochemical pathways.