The Neural Ordinary Differential Equations paper by Chen et. al. was awarded as one of the 4 best papers at NeurIPS 2018. The core idea is to model a time-series by modelling its derivative with a neural network. In Julia you can readily apply this approach, also with stochastic, delayed and other differential equations. In this workshop you are going to implement such state-of-the-art models yourself – with ease, thanks to Julia. We will go over the theory behind neural differential equations, why Julia is especially apt for this, and experience hands-on its power for predicting time-series. Looking forward to see you all! Prerequisites: No prerequisites. Everyone is highly welcome. If you want to join the hands-on part, just bring your own laptop with junolab.org installed.