The recent rapid improvement of AI techniques in general has also had a large impact on the way seismological data can be processed. Several machine learning algorithms determining seismic onset times have been published in the last years facilitating the automatic picking of large data sets. In this study we apply the deep neural network PhaseNet to a network of over 900 permanent as well as temporal broad band stations that have been recording during 2016-2020 as part of the AlpArray research initiative in the Greater Alpine Region. We developed a purely data-driven pre-inversion pick selection method to consistently remove outliers from the automatic pick catalog which allows us to include observations throughout the crustal triplication zone. Using the established VELEST and the recently developed McMC codes we invert for the 1D P- and S-wave velocity structure including station correction terms while simultaneously relocating the 384 seismic events with M_L >= 2.5 selected for this study. As a result we present two separate models differing the the included maximum observation distance and their suggested usage. While the AlpsLocPS model is based on arrivals from < 130km and can be used to consistently (re)-locate seismicity based on P & S observations the GAR1D_PS model includes the entire observable distance range of up to 1000km and for the first time provides consistent P- & S-phase synthetic travel time for the entire Alpine orogen. Comparing our relocated seismicity with hypocentral parameters from other studies in the area we quantify the absolute horizontal and vertical accuracy of event locations as ~2.0km and ~6.0km, respectively. Additionally, we will present preliminary results of our 3D P- & S-wave velocity model for the entire Alpine orogen.