Parallel processing is a major source of computational power in the brain. Accordingly, connectomes of complete circuits uncover many anatomical pathways for information flow. How do different pathways implement different processing? While differences in connectivity are clearly important, differences in synaptic inhibition, short-term synaptic plasticity, and intrinsic cellular biophysics are all likely to contribute as well. However, linking these mechanisms to the circuit-level segregation of computation has been challenging. To overcome these challenges, my laboratory studies parallel processing in Drosophila olfaction, a model system where single second-order projection neurons diverge to target many third-order neurons. We use a combination of in vivo patch-clamp electrophysiology, 2-photon optogenetics, and calcium imaging from neurons with known connectivity to investigate the biophysical properties that enable and constrain signal propagation through divergent networks. I will share recent results demonstrating how olfactory coding in projection neurons is diversified by their downstream targets in the lateral horn and mushroom body, and how this depends on a striking amount of diversity in synaptic, cellular, and circuit properties. Our work is revealing how parallel processing is organized and implemented in the brain and establishes Drosophila olfaction as an invaluable experimental platform for testing its functional roles.