Dagster is a new type of workflow engine, called a data orchestrator, designed to manage complex data systems. It introduces a data-aware, typed, self-describing logical orchestration graph that models the implicit structure in every data application and platform. This graph enables structured programming, testability, reliability, and observability, making it easier for data teams to collaborate, debug, and deploy their systems. The orchestrator targets full-stack engineers, data scientists, and platform teams, providing flexibility and adaptability to run any tool, use any storage, and deploy to any infrastructure. It addresses the pain points of traditional workflow engines by introducing a more structured programming model that exposes a richer, semantically aware graph. By adopting Dagster's API, users can build reusable components, manage complexity, and achieve end-to-end ownership of data production. The system is designed to work with heterogeneous tools, teams, and infrastructure, providing process isolation, deployability, and operational efficiency. It aims to empower data practitioners to become participants in a software engineering process, enabling them to contribute to the development of their own applications and tools.