Studying cosmology with large galaxy surveys requires an unprecedented understanding and mitigation of systematics — a challenge that can be addressed on two fronts: quantification of the impacts of systematics, and new tools to mitigate them. In this thesis, we study the impacts of both observational and astrophysical systematics. Focusing on the Legacy Survey of Space and Time (LSST) to be carried out by the Vera C. Rubin Observatory, we quantify the impacts of LSST observing strategy on large-scale structure studies and demonstrate the effectiveness of large translational dithers in increasing LSST survey uniformity and reducing systematic uncertainties (Awan et al. 2016, ApJ, 829, 50). We also study the impacts of Milky Way dust on dark energy science and demonstrate that ~25% of the default LSST main survey area would not be useful for extragalactic static science given the Milky Way dust extinction, motivating the reconfiguration of the LSST survey footprint to avoid high-extinction regions of the sky (Lochner et al. 2018; Olsen et al. 2018). Then, we present new galaxy angular correlation function estimators that explicitly correct for sample contamination arising from photometric redshift estimation (Awan & Gawiser 2020, ApJ, 890, 78). And finally, we present our work on quantifying the level of some key systematics (observing strategy, seeing, photo-zs, Milky Way dust) that leads to e.g., a 1-sigma bias in our cosmological parameter estimation using 2-point galaxy clustering (Awan et al., in prep). While these techniques are motivated by preparations for LSST, they are applicable to other large galaxy surveys like Dark Energy Survey (DES), Dark Energy Spectroscopic Instrument (DESI), Hobby-Eberly Telescope Dark Energy Experiment (HETDEX), Euclid, and Nancy Grace Roman Space Telescope.