Marketplaces like eBay, AirBnB and Uber require careful design to work well. In this talk, I will provide an overview of our research on how to integrate “machine learning” into a market mechanism to improve efficiency. I will provide two practical examples. First, I will describe the design of a machine learning-powered iterative combinatorial auction which may be used by a government to auction off different 5G licenses among multiple mobile network operators. Second, I will describe our work on the design of a machine learning-powered adoption platform, where the goal is to use a data-driven approach to find compatible matches between children and families. I will conclude by discussing how this new machine learning-powered market design paradigm can also be applied to other problems like school choice or course assignment.