The benefits linked to the flexibility offered by additive manufacturing must not hide the new constraints induced by the industrialization of such a manufacturing process. Specific design rules must be implemented in the design phase, to increase both quality and productivity in the development chain. Thus, some manufacturing constraints could already be formulated and integrated into the topology optimization design problem, such as a minimum resolution induced by the printing, a maximum feature size to limit overheating in the produced parts, the control of overhanging surfaces to limit the use of supports, as well as closed cavities constraints. But others remain to be investigated so that the concepts generated can be directly produced. The present work aims at presenting a coupling strategy between the part topology optimization and the additive manufacturing build phase preparation. Both aspects are implemented in two standalone in-house software packages. On one hand, topology optimization capabilities including XFEM-level set and lattice structures are implemented on top of our thermo-mechanical kernel Morfeo, while on the other hand the rapid geometry analysis of the part including the critical areas for the manufacturing are included in Cafeine. The latter includes many other relevant capabilities: identification of closed cavities (trapped powder), roughness evaluation, recoater impact detection, identification of surfaces requiring the addition of supports and estimation of the volume of supports, identification of the risk of de-powdering in channels (channels too long or too narrow), identification of sudden changes in cross-section that may be difficult to print, evaluation of the manufacturing time according to the orientation of the part on the plate. In this framework, a parametric surrogate-assisted optimization (performed by our in-house optimization platform Minamo) which aims at minimizing the quantity of support as well as the manufacturing time is illustrated on an industrial test case.