Set up Exasol AI Lab

Learn how to set up Exasol AI Lab to work with in-database AI in Exasol.

Exasol AI Lab is a pre-configured Docker container that gives you a ready-to-use Jupyter environment for data science and machine learning on Exasol. It bundles the Notebook Connector, SageMaker Extension, Transformers Extension, and Text AI Extension together with example notebooks, so you can connect to your database and start training, deploying, or running models without setting up the stack yourself.

Watch the following videos for a quick introduction to Exasol AI Lab:

What is included

Exasol AI Lab bundles the following components into a single container:

  • Jupyter Notebook server as the primary development interface
  • Notebook Connector for persistent, encrypted database credential management
  • SageMaker Extension for managed model training on AWS SageMaker
  • Transformers Extension for running Hugging Face models inside Exasol using UDFs
  • Text AI Extension for in-database text analysis (NER, summarization, keyword extraction)
  • Example notebooks covering scikit-learn workflows, SageMaker integration, Hugging Face model deployment, and more

Prerequisites

Before you start, make sure you have:

  • Docker installed and running on your machine – see Get Docker
  • An Exasol database that the container can reach over the network
  • Sufficient system resources – see the Exasol AI Lab Docker usage guide on GitHub

Install and run Exasol AI Lab

  1. Pull and start the Exasol AI Lab container:

    Copy 
    docker run --publish 0.0.0.0:49494:49494 exasol/ai-lab

    This maps port 49494 inside the container to port 49494 on your host machine. The Jupyter server starts automatically.

  2. Open your browser and go to http://localhost:49494

  3. Log in with the default password: ailab

Change the default password after your first login, especially if the container is accessible on a shared network.

By default, data inside the container is not persistent. If you stop or remove the container, any work you saved inside it is lost. To persist data, add a volume mount to the Docker command.

See the Exasol AI Lab Docker usage guide for volume mount options, custom password configuration, and GPU passthrough.

Configure your database connection

Once you are logged into the Jupyter interface:

  1. Open the main_config.ipynb notebook from the file browser.
  2. Run the notebook cells to enter your Exasol database host, port, and credentials.
  3. The Notebook Connector stores your credentials in an encrypted local secret store. You only need to do this once, and all other notebooks in Exasol AI Lab will use the same connection.

First steps in Exasol AI Lab

After configuring your database connection:

  1. Open first_steps.ipynb for an introduction to the Exasol AI Lab environment and its capabilities.

  2. Explore the example notebooks for specific use cases:

    • scikit-learn: Train and deploy classic ML models using your Exasol data
    • SageMaker: Run managed model training on AWS and deploy results back to Exasol
    • Hugging Face: Download transformer models into BucketFS and run inference through SQL UDFs

Each example notebook walks you through the full workflow from data preparation to model deployment.

Next steps