This study combines numerical simulation of fluid flow and heat transfer with machine learning (ML) and evolutionary algorithms to optimize the shape of structured laminar channels for improved heat transfer. The initial phase of the study involves gathering quantitative data on fluid flow and heat transfer in structured channels via numerical simulation, which is then utilized to train a ML model. The trained ML model is observed to predict flow and heat transfer characteristics with improved speed and accuracy compared to traditional simulations. The ML model is then utilized as a surrogate, instead of numerical simulations, with particle swarm optimization (PSO) to explore the high-dimensional parameter space of structured channel geometry for three different objective functions − minimizing C_ƒ , maximizing St and Reynolds analogy (RA) factor. The optimization expectedly delivers a flat channel geometry for minimization of C_ƒ and suggests a complex small-scale meandering geometry - intensifying the wall-normal heat transfer through vortex shedding and flow detachment - for maximizing St. For maximizing RA the optimization proposes a large-scale meandering geometry, which maintains the flow laminar but slowly redirects it from one wall to another. The results indicate that the combination of numerical simulations with ML techniques and evolutionary algorithms can effectively process and distill information from existing data, reducing the need for extensive simulations in the design optimization process.