Arroyo is a distributed stream processing engine built in Rust, designed for stateful computations on both bounded and unbounded data streams. Unlike traditional batch processing, Arroyo processes data in real-time, providing results in less than a second. Key features include SQL support for streaming pipelines, the ability to handle millions of events per second, and stateful operations such as windows and joins. Arroyo also offers state checkpointing to ensure fault tolerance and recovery, and it utilizes the Dataflow model for time-oriented stream processing. The engine supports various connectors, like Kafka and Iceberg, allowing it to integrate easily into existing data ecosystems. Use cases range from fraud detection to real-time analytics and machine learning feature generation. Arroyo aims to differentiate itself from established engines like Apache Flink and Spark Streaming by offering serverless operations that scale seamlessly in cloud environments. The SQL performance is prioritized, and the design is user-friendly, making it accessible for those without deep expertise in streaming technologies. Installation is straightforward, either via Homebrew for MacOS or with a shell script for MacOS and Linux. Users can also run Arroyo in Docker, and a Web UI is available for monitoring. For those who prefer not to self-host, Arroyo is available as a managed service on the Cloudflare Developer Platform, currently supporting stateless pipelines. The project encourages community contributions and offers several channels for support and feedback.