Have Docker and Docker Compose installed on your machine.

1. Copy or create a .env file in the project root:

MYSQL_ROOT_PASSWORD=pizzatime
MYSQL_DATABASE=daver_db
MYSQL_USER=daver
MYSQL_PASSWORD=pizzatime

There are 2 options for setting up the AI models:

1. Using Ollama (Recommended because the models run way faster):

   • Install Ollama on your machine.
   • Download the required models using:

     ollama pull phi4-mini:3.8b
     ollama pull gemma3:4b
     ollama pull deepseek-coder:6.7b

   In the .env file, set the OLLAMA_HOST variable to point to your Ollama instance:

       OLLAMA_HOST=http://host.docker.internal:11434

2. Using Docker:

   • If you prefer to run the models in Docker, you can use the provided docker-compose.yml file.
   • Make sure you allocate enough resources to the models in your Docker settings. Should be at least 16GB of RAM.
   • The models will be pulled automatically when you start the project.
   • We recommend using Ollama locally instead of Docker for better performance.

./start-daver.sh # or
./start-daver.sh -d # for detached mode

A database with some sample data will be created automatically. The test data was obtained from the databasestar sample databases (https://github.com/bbrumm/databasestar/tree/main/sample_databases/sample_db_olympics)

To test the application, you can use the daver testing UI: https://github.com/boehmmatthias/daver-ui

Start the testing UI like mentioned in the projects README.md. After that you can access the application at http://localhost:8080 in your web browser. The application provides a simple interface to interact with the database using natural language queries.

For the database config file you can use the daver_sample_config.yaml file in the root directory.

The database config file should look like this:

database:
  host: "db"
  port: 3306
  username: "daver"
  password: "pizzatime"
  database: "daver_db"