This talk provides an overview of how inverse optimization can be applied in radiation therapy (RT) treatment planning. The idea is to reverse engineer "good" treatment parameters from historical treatments so as to learn appropriate objective functions and weights. Combined with machine learning approaches, inverse optimization can reduce the need for manual trial-and-error planning iterations, improving planning efficiency while maintaining personalization of treatments. In consideration of the theme of the workshop, the talk ends with a short tribute on robust optimization in radiation therapy.