The rapid development of autonomous vehicles (AVs) holds vast potential for transportation systems through improved safety, efficiency, and access to mobility. However, due to numerous technical, political, and human factors challenges, new methodologies are needed to design vehicles and transportation systems for these positive outcomes. This article tackles technical challenges arising from the partial adoption of autonomy: partial control, partial observation, complex multi-vehicle interactions, and the sheer variety of traffic settings represented by real-world networks. The article presents a modular learning framework which leverages deep Reinforcement Learning methods to address complex traffic dynamics. Modules are composed to capture common traffic phenomena (traffic jams, lane changing, intersections). Learned control laws are found to exceed human driving performance by at least 40% with only 5-10% adoption of AVs. In partially-observed single-lane traffic, a small neural network control law can eliminate stop-and-go traffic -- surpassing all known model-based controllers, achieving near-optimal performance, and generalizing to out-of-distribution traffic densities.