Distributionally robust average-reward reinforcement learning studies sequential decision-making problems in which the objective is to maximize long-run average performance under uncertainty about the underlying transition dynamics or reward distributions. While distributionally robust reinforcement learning has been studied primarily in discounted and finite-horizon settings, its theoretical foundations and computational methods under the average-reward criterion remain relatively underdeveloped. Nevertheless, this formulation is particularly relevant for continuing systems that operate indefinitely and face model misspecification or distributional shifts, including queueing systems, inventory control, supply chains, and online platforms. In this project, students will learn the necessary background in average-reward Markov decision processes and distributionally robust optimization, investigate the statistical and structural properties of robust average-reward policies, and develop scalable algorithms supported by numerical experiments. Due to the technical complexity of the project, advanced proficiency in real analysis and Markov chains is required.