Serverless platforms promise to enable users to focus on application logic instead of infrastructure, thereby improving developer productivity. Current offerings obviate the need to manage resources and configuration by instead presenting the user with a limited set of preset resource allocation configurations (that couple memory and CPU resources) from which to choose. However, Bilal et al. find that these configurations do not typically achieve optimal performance. They propose using Bayesian Optimization algorithms to automatically allocate resources (i.e. to find the best combination of CPU share, memory limit, and VM type) for a user’s functions or present the users with the best set of configuration knobs based on their preferences and requirements.