PostgreSQL for Data Science & AI Workloads

This repository provides a Docker image for PostgreSQL, pre-configured with powerful extensions tailored for data science, analytics, and AI/ML workloads.
The image is built in two stages:

1. Building PostgreSQL Extensions (e.g., pg_parquet from CrunchyData).
2. Final PostgreSQL Image with compiled extensions and additional PostgreSQL packages.

---

🚀 Why Use This Image?

PostgreSQL is a powerful data processing engine, and with the right extensions, it can serve as a feature store for AI/ML models, an analytics database, and a scalable data warehouse.
This image includes:

• Vector search for AI applications (pgvector)
• Time-series analysis for real-time data (TimescaleDB)
• Efficient JSON querying for structured/unstructured data (jsquery)
• Partitioning for large datasets (pg_partman)
• Query optimization & hinting (pg-hint-plan)
• Advanced auditing for compliance/security (pgaudit)
• Job scheduling inside PostgreSQL (pg_cron)
• Graph database support for network analysis (AGE)
• Event queueing system for real-time data streaming (pgq3)
• Advanced regex support with Perl-compatible regex (pgpcre)
• Built-in Parquet file format support (pg_parquet)
  ⚠️ Note: Building pg_parquet is resource-intensive and may take a long time.

---

📦 Features & Pre-installed Extensions

• PostgreSQL
• Optimized for AI, data science, and analytics
• Pre-installed PostgreSQL extensions:
  • pg_parquet (Parquet file format support)
  • postgresql-17-age (Graph database support)
  • postgresql-17-cron (Job scheduling within PostgreSQL)
  • postgresql-17-jsquery (Advanced JSON querying)
  • postgresql-17-partman (Automated table partitioning)
  • postgresql-17-pg-hint-plan (Query optimizer hints)
  • postgresql-17-pgaudit & postgresql-17-pgauditlogtofile (Audit logging)
  • postgresql-17-pgpcre (Perl-compatible regular expressions)
  • postgresql-17-pgq3 (Queueing system for PostgreSQL)
  • postgresql-17-pgvector (Vector search for AI/ML workloads)
  • postgresql-17-prefix (Prefix search optimization)
  • postgresql-17-timescaledb (Time-series database functionality)
• Supports Rust-based PostgreSQL extensions via pgrx.
• Built-in health check (pg_isready).

---

🏗️ Building the Image

Pull from Docker Hub

docker pull rbehzadan/postgres

Build Locally

To manually build the image:

docker build -t my-postgres .

⚠️ Note: Building pg_parquet requires significant CPU and memory resources, as it compiles from source.

---

📦 Running the Container

Start a PostgreSQL container with this image:

docker run -d --name postgres \
  -e POSTGRES_USER=admin \
  -e POSTGRES_PASSWORD=secret \
  -e POSTGRES_DB=mydb \
  -p 5432:5432 \
  rbehzadan/postgres

Connect to the database using psql:

docker exec -it postgres psql -U admin -d mydb

---

🛠️ Enabling Extensions

Once connected, enable any installed extension:

CREATE EXTENSION pg_parquet;
CREATE EXTENSION timescaledb;
CREATE EXTENSION pgvector;
CREATE EXTENSION age;
CREATE EXTENSION pgq3;
CREATE EXTENSION pgpcre;

---

🔍 Health Check

Verify the container's health status:

docker inspect --format='{{json .State.Health}}' postgres

---

🤝 Contributing

If you find a bug, want to add an extension, or improve the build process, feel free to open an issue or submit a pull request.

---

📜 License

This project is licensed under the MIT License.