There is currently a large spatiotemporal gap between experimentally obtained defect-engineered crystals and the ab initio modeling of such structures at the mesoscale. In this project, we focus on the Metal-Organic Framework UiO-66 and its catalytic properties. Very recently, we were successful at creating well-defined and unique defects in this crystal structure and to link this to catalytic properties. On the other hand, theoretical modeling is still either quantum-accurate on the small scale (<10 nm, DFT) or largely descriptive on the mesoscale (50 nm, Force Fields). In this project we will take several hurdles at the same time: 1) we will finetune and fundamentally understand the mechanisms that lead to these well-defined defects, using several synthetic approaches, recently developed by us; 2) we will develop new methodologies to model defective crystals up to the 50 nm-scale with quantum-accuracy, using Machine Learning Potentials, which shows a lot of potential for MOFs. The defective crystals will be experimentally and theoretically validated using three judiciously chosen catalytic reactions to further understand the catalytic reaction cycle and the effect of defects (on the mesoscale) on the catalytic mechanism.