Arroyo is now open-source! Read the announcement.

Transform, filter, aggregate, and join Kafka streams using SQL, with sub-second results. Automatically scale from 10 events to millions per second. Pay only for what you use.

Serverless stream processing

A screenshot showing the Arroyo user interface
Backed by

Why Arroyo

Fully serverless

No clusters to manage. No fixed costs. Effortless autoscaling.


You already know how to build streaming pipelines. Easily upgrade batch jobs to real-time.

Works with your data infra

Read from your existing Kafka or Kinesis streams. Write to one of dozens of sinks.

Three Steps to Streaming

1. Connect your stream

Tell us how to connect and authenticate to your Kafka or Kinesis streams. We support Confluent cloud, MSK, self-hosted Kafka, and data in JSON, Protobuf, or Avro.

2. Write SQL

Construct your pipelines using the SQL you already know, plus streaming window functions and WASM user-defined functions.

3. Connect your sinks

Tell us where to send your data. We support stream sinks and many other data systems, or query your results directly via our API.

What can you build with Arroyo?

On-demand businesses

Traffic accidents. Supply-demand imbalances. Unsafe situations. If you operate in the real world, you need to incorporate information in seconds.

Real-time analytics

No need to wait for the daily batch job to know how your business is doing. With Arroyo, you can move your analytics to real-time without added complexity or cost.

Machine Learning

Generate ML features in real-time to proactively respond to the changing world and customer behavior.

Alerting and Risk

Transform logs into metrics and find anomalous behavior of your systems. Catch fraudsters before they exploit your business.


A headshot of Micah Wylde
Micah Wylde

Micah was previously tech lead for streaming compute at Splunk and Lyft, where he built real-time data infra powering Lyft's dynamic pricing, ETA, and safety features. He spends his time rock climbing, playing music, and bringing real-time data to companies that can't hire a streaming infra team.

A headshot of Jackson Newhouse
Jackson Newhouse

Jackson spent a decade at Quantcast building in-house distributed systems. He relishes designing maximally efficient systems with aggressive performance targets and massive scale. He's thrilled to be helping companies move their data processing into the stream-first future.