This repository contains a simple multi-modal example of PostgreSQL, showcasing various extensions and functionalities, including AI capabilities.

In addition to powerful extensions to support flexibly shaping and manipulating data, it also includes the PostgresML plugin, which elevates PostgreSQL to a ready-to-go AI platform. With PostgresML, you can seamlessly integrate artificial intelligence and machine learning capabilities into your PostgreSQL database.

By combining the versatility of multi-modal data management with the power of AI, this repository provides you with a solid foundation for building intelligent data applications. Whether you are working on advanced analytics, predictive modeling, or complex machine learning tasks, PostgreSQL with PostgresML empowers you to leverage the data stored in the database and extract valuable insights.

Feel free to explore, experiment, and extend the capabilities of PostgreSQL with AI using this repository. Join us in unlocking the potential of AI-driven data management and analytics with PostgreSQL's flexibility and PostgresML's intelligence. Let's embark on a journey of innovation, where data meets AI and possibilities are limitless.

The "postgres-multimodal" project, including PostgresML, showcases the incredible versatility of PostgreSQL as a multi-modal database and AI platform. In today's evolving data landscape, databases must adapt and support various architectural types to enable powerful AI-driven data analytics.

The project's purpose is to demonstrate that PostgreSQL is not limited to traditional relational data but can also handle geospatial information, time-series data, graph structures, and more. With the addition of PostgresML, it becomes a comprehensive AI platform that empowers users to perform advanced analytics and machine learning directly within the database environment.

Why is this important, you ask? Well, the future of data management and analytics is not limited to a single data model or separate AI platforms. With the rise of IoT devices, geolocation-based applications, social networks, and complex interconnected systems, our data has become more diverse and interconnected than ever before. By combining the power of a multi-modal database like PostgreSQL with AI capabilities, we unlock new opportunities for businesses to gain valuable insights, make data-driven decisions, and drive innovation.

By having a multi-modal database like PostgreSQL, we simplify data integration, reduce duplication, and make querying different data types a breeze. Plus, with the integrated AI capabilities of PostgresML, we can perform advanced analytics, predictive modeling, and complex machine learning tasks directly on the data stored in the database. This eliminates the need for separate AI platforms and streamlines the workflow, resulting in enhanced efficiency and faster time-to-insight.

So, join us on this exciting journey as we leverage the power of PostgreSQL and PostgresML to enable powerful AI-driven data analytics. Together, we'll unlock the full potential of data, drive innovation, and shape the future of intelligent data management and analytics. Get ready to embark on a transformative experience where the boundaries of traditional databases and AI platforms blur, opening up a new era of possibilities for businesses and AI enthusiasts alike.

• docker-compose.yaml: The Docker Compose configuration file that sets up the PostgreSQL container and runs the Liquibase container to migrate.
• Dockerfile: The Dockerfile used to build the PostgreSQL container. It installs Apache Age and sets up the necessary configurations to allow non super users to interact with it.
• migrations/root.changelog.xml: The Liquibase changelog file that includes SQL scripts for database migrations.
• migrations/sql/00010_initial_database_setup.sql: SQL script for initializing the database with various extensions and configurations.

To play with 'postgres-multimodal' locally, follow these steps:

1. Make sure you have Docker installed on your system with the docker compose extension.
2. Clone this repository: git clone https://github.com/your-username/postgres-multimodal.git.
3. Navigate to the cloned directory: cd postgres-multimodal.
4. Start the Docker containers: docker compose up -d.
5. Wait for the containers to build and be up and running. PostgreSQL will be available on port 5432.

You may then connect to the PostgreSQL database using your preferred client or command-line tool.

• Host: localhost
• Port: 5432
• Database: postgres
• Username: postgres
• Password: password

Alternatively, you may connect to the database using the running postgres-multimodal docker container.

docker exec -it postgres-multimodal-postgres-multimodal-1 psql

Once connected, you can interact with the database and explore the various extensions and functionalities provided by this multi-modal example.

The examples directory contains example usage.

The PostgreSQL image built in this example includes the following extensions to enable multi-modal behavior:

• postgresml: Enables AI workloads at the database level.
• postgis: Enables geospatial capabilities.
• timescaledb: Provides support for time-series data.
• pgvector: Supports vector-based operations used for similarity and semantic search.
• apache age: Enables graph and Cypher functionality used to enable graph database functionality found in database like Neo4j.
• hll: Enables the HyperLogLog extension, used to efficiently count large amounts of distinct values.

Several other minor extensions are also enabled by default.

• uuid-ossp: Allows generating UUIDs.
• pg_trgm: Offers trigram-based similarity matching.
• pgcrypto: Provides cryptographic functions.

Liquibase is used to run the database migrations specified in the root.changelog.xml file against the PostgreSQL container. The migrations/sql/00010_initial_database_setup.sql file contains SQL statements for initializing the database with extensions and configurations. Additional migration files may be added to the migrations/sql directory to extend the behavior.

To leverage the graph features of Apache AGE, the following two commands must be run per session.

-- Required
LOAD 'age';

-- Optional, but recommended
SET search_path = ag_catalog, "$user", public;

For more information on Apache AGE, see the official Apache Age Documentation.

The multi-modal image is based off of the timescaledb-ha docker image, which contains all of the extensions listed in the 'Features' section besides Apache Age and PostgresML. This example is theoretically compatible with the timescaledb high-availability helm chart, although this has not been tested.

The repository also includes an example Liquibase changelog file to demonstrate how to configure Liquibase to manage database migrations using SQL files found in a local directory. This allows you to easily version and control your database schema changes while integrating AI capabilities through PostgresML.

Contributions to this repository are welcome. If you encounter any issues or have suggestions for improvements, please open an issue or submit a pull request.

This repository is licensed under the Apache 2.0 License.