This document contains slides from a presentation on building resilient serverless applications. Some key points discussed include: designing applications to withstand failures through techniques like running in multiple availability zones and regions; handling failures from overloading, distributed systems, and timeouts; implementing patterns like decoupled invocation and sagas for distributed transactions; dealing with partial failures, retry storms, and poison messages; and performing chaos engineering experiments to identify weaknesses. The overall message is that failures will occur, so applications need to be designed to recover quickly from difficulties through resilience.