I will give an overview of auxiliary-field quantum Monte Carlo (AFQMC), with the goal of making connections with other methods of
interest at this workshop. AFQMC embodies almost all the major issues in Monte Carlo methods for quantum systems; it also can be thought of quite naturally in the framework of neural networks. I will discuss opportunities for extending the reach of current capabilities, and highlight a few recent algorithmic advances in ab initio simulations in solids, including correlated sampling, the computation of gradients (forces and stresses), and structural optimization.