Data engineering in python

16-09-2024

General tools

• compute
  • Databricks (which uses Apache Spark as it's processing engine and allows for in-memory caching and optimised query execution) and DeltaLake (which provides the data lake / warehouse)
• storage
  • relational (MSSQL, PostgreSQL)
  • non-relational (MongoDB, Redis)
  • object (Azure blob / Amazon S3)
• containerisation
  • Docker (automate and manage the deployment of applications inside containers)

Python tools

• Airflow (task-centric, great open source support) and Dagster (asset-centric, great testing and debugging support) for orchestration
  • https://github.com/apache/airflow
  • https://github.com/dagster-io/dagster
• Polars and Pandas for data processing (Polars is becomming more mainstream, it uses Rust and Apache Arrow on the backend)
  • https://github.com/pola-rs/polars
• DBT for data transformations
  • https://github.com/dbt-labs/dbt-core
• FastAPI for building API's (this is a very popular lilbrary, and works on top of Pydantic which handles data parsing and validation)
  • https://github.com/fastapi/fastapi
  • https://github.com/pydantic/pydantic
• SQLAlchemy for database connections (they released a new 2.0+ major release a few months ago)
  • https://github.com/sqlalchemy/sqlalchemy
• Poetry / UV for project, package and dependency management
  • https://github.com/python-poetry/poetry
  • https://github.com/astral-sh/uv
• Ansible for infrastructure management and configuration
  • https://github.com/ansible/ansible
• Opentelemetry for monitoring (open source, language agnostic, can capture rich metadata)
  • https://opentelemetry.io/