Wilem Röpke recently obtained his PhD with the thesis Thinking in Trade-Offs: Training Agents to Balance Conflicting Objectives under Uncertainty. While life constantly requires us to trade-off multiple goals – speed vs safety, cost vs quality, time vs money – many AI algorithms are built to optimise a single objective. In his work, Röpke showed how existing AI techniques can be adapted to deal with multiple objectives. He developed theoretically strong and practically useful ways of taking into account user preferences regarding trade-offs and investigated how multiple single-objective AI-systems can be used to tackle complex multi-objective problems.