Getting Started


There are two main flows for using Aquarium:
As illustrated above, you'll need to do the following:
  1. 1.
    Upload your data (Python client): Aquarium provides a Python client library that makes it easy for you to upload your datasets + model inferences, so that they can be viewed on our web app. To set up your own upload script, see our docs for more details.
  2. 2.
    Explore your data (web app): Once your ground truth dataset and inference are uploaded, you'll be able to explore them in our web app ( To understand all the different ways to investigate your data and triage issues, see our docs for more details.
What if I want to see the Aquarium tool without uploading my data? Note that it's possible to play around with Aquarium on some public demo datasets without needing to upload your own data first. If you'd like to use Aquarium on your own data, fill out this form and we can reach out to set you up with a demo account.


You can join our Slack channel here to ask questions, talk about ML projects, and meet other people in the community. If this link doesn't work, just email [email protected]
Copy link