Speaker: Andrew Green Abstract: XVAs models are amongst the most computationally intensive in finance and require the use of acceleration techniques such as GPU computation and Adjoint Algorithmic Differentiation (AAD). Deep learning provides a computationally efficient and implementation friendly way to approximate derivative valuation function, a critical component of XVA models. This presentations shows how deep learning dovetails with other traditional quantitative finance models to deliver an effective XVA calculation platform.