Materials and structures with tailored nonlinear mechanical responses enable the development of innovative technologies such as soft robots, wearable devices, advanced footwear, and energy-absorbing systems. However, realizing such designs remains a significant challenge. Moreover, there is a growing demand for reprogrammable architectures capable of adapting to multiple tasks. In this talk, we introduce an inverse-design framework for discovering flexible mechanical metamaterials with specific nonlinear responses. By encoding the desired functionality into the design process, this framework optimally tunes the metamaterial's full-scale geometry using a fully differentiable simulation environment. Through this strategy, we design mechanical metamaterials tailored for tasks such as energy focusing, energy splitting, dynamic protection, and nonlinear motion conversion.