
Learn what this course covers and what you will build. You will get an overview of MPP databases, Exasol Personal, UDFs, the AI Lab, and the conversational BI setup with MCP.
Understand the core idea behind massively parallel processing and how it differs from traditional relational databases. Learn why Exasol stands out compared to managed cloud warehouses like Snowflake and BigQuery in terms of cost, speed, and data sovereignty.
Set up an IAM user, configure the AWS CLI, and deploy a fully functional Exasol cluster on your own EC2 instance using the Exasol Launcher. By the end of this lesson you have a free enterprise-grade database running in your own AWS account.
Connect to your Exasol cluster using DB Visualizer, the admin UI, and the CLI. You will create a schema and table and import 48 million rows of real NYC taxi data from S3.
Learn how Exasol stores data in compressed columnar format in memory and why that makes queries so fast. Understand how distribution keys and partition keys work and how to choose them correctly from the start.
See with real numbers what adding nodes does to import speed and query performance. Run the same query on a badly distributed and a well-distributed table to understand the impact of your data layout decisions.
Learn how to write Python UDFs that run directly inside Exasol and can be called from SQL. You will build a UDF that calls a live financial API and combines the results with data already in your database, all in a single query.
Set up the Dockerized Jupyter environment that enables machine learning workflows against your Exasol cluster. You will configure BucketFS credentials and activate the script language container needed for the next lessons.
Train a decision tree classifier on a credit application dataset inside the AI Lab and upload the model to BucketFS. Then run predictions on new data directly from a SQL query without any external Python environment.
Load a zero-shot classification model from Hugging Face into BucketFS and classify text documents directly from SQL. No data leaves the database and no external API is called.
See how an MCP server combined with a locally running LLM lets you query your Exasol database in plain English. Learn how the architecture works and why everything stays on your own infrastructure.
A recap of everything covered in the course, from MPP fundamentals to in-database AI and conversational BI. You will also get guidance on where to go next and how to apply what you learned to your own data.
Most data teams treat the database as a place to store and query data. This course shows you a different way of thinking about it.
Exasol is a massively parallel processing database that keeps your data compressed in memory and spreads the workload across nodes. That alone makes it significantly faster than traditional data warehouses. But what makes it truly powerful is what you can run inside it: Python code, machine learning models, and even language models, all called directly from SQL.
In this course you will set up a fully functional Exasol cluster on your own AWS infrastructure using Exasol Personal, which is completely free and has no feature or data limits. You will load real datasets, design tables with the right distribution and partition keys, and see with real benchmark numbers what proper data layout does to query performance.
From there you will write Python UDFs that call live APIs and return results inside a single SQL query. Then you will move into the AI Lab, where you will train a scikit-learn classifier, upload it to BucketFS, and run predictions straight from SQL. You will also load a Hugging Face zero-shot classification model into the database and classify text documents without sending a single row to an external service.
By the end of this course you will know how to turn Exasol into a full in-database AI and ML platform that runs entirely on your own infrastructure.