Thrombosis on synthetic biomaterial surfaces hinders the development of blood-wetted artificial organs. One neglected source of blood clot in left ventricular assist device (LVAD) is the inflow cannula, where emboli are produced. The emboli may be later ingested and ejected from the pump. A widely explored strategy to prevent surface-induced thrombosis in cardiovascular implants is to promote stable endothelium, which is the natural thrombo-resistant blood-contacting surfaces, and is deemed to be the only long-term solution for hemocompatible materials. To induce endothelialization and improve hemocompatibility, LVAD manufacturers adopted powder sintering in multiple contemporary LVADs. Despite the wide adoption of sintered titanium, clinical outcomes over the past several decades have been highly variable, ranging from neointima to thrombus. Chapter 2 addresses this ambiguity by introducing a unified taxonomy for biological response to sintered titanium in LVADs, distinguishing neointima---with anti-thrombogenic endothelium--- from pannus, pseudoneointima, and thrombus. In Chapter 3, I studied platelet deposition, which is the first step of blood interacting with the biomaterial. Using a microfluidic platform mimicking the HeartMate 3 cannula’s surface, platelet deposition was visualized in real time. Experimental and computational results demonstrated that lower shear rates paradoxically increased deposition, while embolic events—quantified via an entity-tracking algorithm—varied spatially. This result suggests that the sintered surface of the ventricular cannula may engender unstable thrombi with a greater likelihood of embolization at supraphysiological shear rates. In Chapter 4, I design microtrenches to regulate local wall shear stress, which enables endothelialization under supraphysiological shear. An auto-optimization strategy was implemented with an objective function measuring area coverage of endothelial-favorable WSS. A maximum of 79% linear coverage was achieved under supraphysiological shear stress, a significant improvement over conventional designs (30%). This study offers a transformable approach to design hemocompatible surface modification that could promote EC retention in high-shear implants. Chapter 5 advanced a multi-constituent thrombosis model that integrates endothelial nitric oxide (NO) dynamics, which a critical endothelial release anti-thrombotic signal. This model introduces shear-dependent NO production and transportation, which is qualitatively validated against previous experimental observation. By incorporating NO in thrombosis simulation, the simulation of platelet deposition on endothelial covered microtrenches, which is optimized in the previous chapter, reduced drastically. The results highlight its potential to predict endothelial-mediated hemocompatibility and guide future biomaterial topography design. Finally, I suggest future directions in translating platelet deposition insights to 3D sintered geometries and validating the thrombosis model through direct measurement of NO gradients and platelet dynamics in vitro under various flow conditions.