Accessing and working with data is not easy with serverless, because traditional methods of database access don't work well. In this talk, I will discuss the key problems around working with databases when building serverless application logic and approaches to solving them. I will motivate problems like connection pooling, cold-starts, handling spiky concurrent loads, database transactions and transient failures. I will then present GraphQL as a possible solution for building a high-performance data API that can scale to serverless workloads. I will talk about the pros & cons of this approach. Finally, I will do some live code demos and make the problems and solutions discussed previously more concrete!