Biological brains are the byproduct of over 500 million years of evolutionary pressures, blindly optimizing for the efficiency and efficacy required to survive our competitive environment. Deep learning aims to build artificial agents that navigate these same environments using neural networks. The field of connectomics provides comprehensive mappings of neural connections, giving insights into suitable neural architectures for well-defined tasks. Currently, there remains a vast difference between the construction of artificial neural networks (ANNs) and biological brains at the level of connectivity and desirable traits, such as inference speed and sample complexity. This gap could be bridged by integrating connectomic data into ANNs, with the hope of recovering these desirable traits. This project explores using architectures derived from fly connectome data in machine learning settings, focusing on identifying the inductive biases that result from these natural structures.