Morphing structures can change their shape in response to thermal excitation and are commonly used to enhance the performance of actuators, soft robotics, and biomedical devices. One of the main challenges is the inverse design problem, in which a deformed target shape is prescribed and the initial material properties, distribution, and geometry are to be determined. To tackle this challenge, a model for the thermo-mechanical response of free standing multi-layered struts that experience moderate to large deflections is developed. The model is validated through thermally activated 3D-printed bi-layer beams. Next, the model is exploited to develop an efficient inverse design algorithm that accepts a target function as an input and computes the referential configuration that can achieve the required thermally-activated deformations. It is shown that the solution to the inverse problem is not unique, thus providing flexibility in design. The merit of the algorithm is illustrated through the inverse design of three thermally-activated strut-based target shapes. The simplicity and robustness of the algorithm reveals fundamental principle guidelines for inverse design of thermally activated shape morphing structures and enables its extension to other stimuli.