We are maintaining an open source repository of tools for censored time-to-event regression. A prospective student would work on various aspects of software engineering and maintenance for the package. Tasks would include torch re implementations of some of the existing tools available in the package in order to make it completely stand alone. We will build on existing python implementations from packages lifelines and scikit-survival, and determine empirically as well as conceptually what the most efficient implementations are. We will then attempt to decide the right software engineering designs to reimplement these methods in a robust fashion in torch. We have access to a multitude of health datasets, including oncology and cardiovascular health in both observational and randomized settings. The student would get a chance to benchmark their implementations against multiple real-world datasets and existing methods. There is also opportunity for students to work with data involving complex modalities, like text and images.