Production data science at your fingertips

Aqueduct abstracts away unnecessarily complex engineering infrastructure and lets you focus on meeting your goals with data science.

Eliminate MLOps Complexity

Spend less time searching StackOverflow and more time building great models

Existing MLOps tools are complex, and they're not designed for data teams. Aqueduct is designed for data teams from the ground up and abstracts away infrastructure and software engineering so that you can focus on data science and machine learning.

When you're ready to take your work to production, Aqueduct's there to make the process painless.

A Simple Python API

Get a production-ready pipeline running in a few lines of vanilla Python. No more worrying about writing a new YAML config every time you want to change a line of code.

Flexible Environments

Run your pipelines anywhere, locally or in the cloud. You can deploy & update your models on your own infrastructure without building custom Docker containers, managing Kubernetes deployments, or storing passwords in plaintext.

Data Integrations

With out-of-the-box connectors to common data systems (Snowflake, Redshift, Postgres, etc.), you can access the freshest data easily & reliably and make sure you're already delivering the latest predictions.

Custom Monitoring

Aqueduct's Checks and Metrics allow you to define measurements & constraints on your workflows, so you can quickly debug and fix errors.

Try Aqueduct

GitHub

Building on a decade of machine learning research

At UC Berkeley's RISE Lab, we spent most of the last decade thinking about how to improve machine learning infrastructure and abstract away low-level cloud tools. Aqueduct builds on that research to make it accessible for every data team.

Given the growing demand for data science in recent years, it's become clear that the existing infrastructure is simply too complicated for most teams to use.

In building Aqueduct, we've focused on how to make data science teams as productive as possible. We believe the core of that mission is building better tools and APIs that abstract away Docker containers and Kubernetes pods to let you focus on what matters.

Instead, Aqueduct works wherever you write your Python and lets you get to the cloud in a couple Python function calls. If that sounds interesting, check out what we're building and say hello!

See how Aqueduct can help your data team

Check out our Quickstart Guide

Guide
Fully open-source
Designed to abstract away underlying infrastructure concerns
Integrated with your (data and compute) infrastructure
Quickly discover bugs, triage errors, and deploy fixes

Stay up to date with the latest on Aqueduct.

Subscribe to our newsletter

Subscribe