Data-centric AI for reinforcement learning

Can you highlight a small example of how Data-centric AI would apply to reinforcement learning type problems