Stars are born in clusters and massive stars die relatively quickly, so we expect core collapse supernovae (SNe) to be clustered in both space in time. Despite this, traditional SN feedback models assume SN blasts are isolated and do not interact. In my thesis I show that clustering might have a very large effect on the final momentum added into the SN's host galaxy's interstellar medium (ISM; as large as an order of magnitude increase), but also that the actual effect size is still unknown and requires further study.

Additionally, I perform experiments into machine learning-driven approaches to identifying nearby dwarf galaxies (a 300:1 class imbalanced problem) within a large (>100 TB) dataset of galaxy images. I show that our current training data and tools do a good start, but are ultimately too limited to get to satisfactory levels of completeness and purity for immediate use within a weak lensing analysis.