Proteins play many essential roles in the human body; muscle contraction, transportation of oxygen in the blood and nerve signals in the brain all depend on them. We can improve our understanding of such processes, and potentially develop treatments if they go wrong, if we know the structure of the proteins involved. To properly see all the details of the protein structures, however, we need to zoom in to such a small scale that we cannot use normal light microscopy. Instead, we have to use other tools. Some of these tools are computational and use physics-based equations and/or machine learning to predict the protein structures. These computational methods can often provide much more details of the protein than experimental methods, e.g. by capturing a "movie" of how the proteins move in space, instead of still images. In my research, I use computer simulations of the proteins and AI-driven tools to determine protein structures. I both use established computational methods to better understand the structure of proteins where we lack the full structural picture, and also try to develop new methods for doing this.