A catalyst's efficiency for accelerating a reaction rate is determined by its molecular structure and interactions with the substrate. While one can predict kinetics for a particular molecular model, tuning the (potentially many) model parameters to reach a desired or optimal kinetics for a catalytic cycle is usually considered computationally prohibitively expensive, especially in solvated systems. Here, we show for a simple model representing a minimal catalytic cycle that such optimization is possible using only one single (path-sampling) simulation, by applying a maximum caliber based path reweighting method. We compute the path ensemble for a single parameter setting of the molecular interactions and then expand the kinetic landscape around these parameters. We find that optimal catalytic turnover or efficiency is orders of magnitude improved and is achieved by relevant parameters that induce strain in the system. Thus, path-reweighting based optimization is not only capable of finding important ingredients that lead to desired kinetic rates but can also identify the mechanistic origins of the rate optimization at a fraction of the costs of a direct evaluation. We demonstrate the versatility of the methodology on a minimal model for kinase signaling. The approach promises efficient computational design of (complex) catalysts using realistic models.