Mitral Valve (MV) regurgitation (MR) is a deadly disease affecting 2% of the world population. Various MV repair techniques such as robotic and transcatheter edge-to-edge repair can provide minimally invasive MR treatment. While these procedures succeed in reducing MR, long term post-repair MR recurrence remains significant, and its causes poorly understood. To guide the clinical decision-making process for MR repair, we have developed a JAX-based Finite Element approach, CARDIAX-MV, to pre-operatively simulate patient-specific MV repair procedures. This new platform greatly accelerates simulations through GPU-based parallelization and provides a differentiable framework for determining patient-specific MV mechanical states. To produce patient-specific reconstructions of the MV, 3D spline-based geometries are fit to an optimal subset of points from a Transoesophageal Echocardiogram. Isogeometric Analysis (IGA) is leveraged in CARDIAX-MV to streamline the image to model pipeline. Another major feature of CARDIAX-MV is the use of Incremental Potential Contact, which provides a robust and high-speed approach for preventing nonphysiological leaflet penetration during MV closure [1]. The main goal of this work is to develop a single integrated high speed computational pipeline to create a full 3D patient-specific predictive MV model. This approach includes determination of MV leaflet and chordae tendineae (CT) mechanical behaviours using an adjoint-based method. This results in a fully calibrated MV model, which is a prerequisite for predicting patient-specific repair outcomes. Current results show CARDIAX-MV can dramatically accelerate inverse modeling of organ-level function compared to our extant commercial code pipelines (ABAQUS). Speedups of 24x, using 5x more data, were observed [2]. Similar speed improvements are expected when applying the inverse modeling pipeline to the determination of patient-specific referential states, which will also be of higher fidelity than existing MV modeling pipelines. This will be useful to cardiologists to understand the mechanisms behind in-silico predictions of surgical repair outcomes.