Optimizing Dimuon Reconstruction in SpinQuest

The SpinQuest experiment at Fermilab presents an excellent opportunity to apply machine learning methods for enhanced data analysis. SpinQuest aims to measure the correlation between the angular distribution of final-state dimuons and the transverse polarization of the target. Accurate reconstruction of the dimuons' four-momentum, originating from the target, is crucial for this measurement. To achieve this, the University of Virginia has developed an innovative approach to reconstruction called Q-Tracker, using a series of deep neural networks. The commissioning and early production runs of SpinQuest taken in the Spring of 2024 offered first-hand experience of both Q-Tracker's online and offline reconstruction abilities. In this talk, we will present the implementation of Q-Tracker for SpinQuest, and review analysis for online and offline reconstruction with the ambition to extract multiple polarized Transverse Momentum Distributions with future data.