Python Best Practices

Python Best Practices developed by Dmitry Mugtasimov during the course of his 27-years career in software engineering.

You can contact me via:

• Email: dmugtasimov@gmail.com
• X: @dmugtasimov
• Telegram: @dmugtasimov

DISCLAIMER: This repository documents Python best practices as I have developed and refined them over 27 years of professional software engineering. They reflect my experience, reasoning, and approach to building and maintaining software, and are shared as a practical reference rather than as a definitive or universal standard. As with any engineering guidance, their applicability may vary depending on context, constraints, and goals.

DISCLAIMER 2: This is "work in progress" repository. If you did not find something you might have expected then it is just not written yet.

DISCLAIMER 3: There are a lot of changes happened since the beginning of 2026 due to the AI advancements. Not all them have been reflected here yet.

Table of Contents

• Technology Stack
• General Principles
• General Coding Rules
• Python Specific Rules

Technology Stack

Core

• Python - programming language
• Django - web framework
• Django REST Framework (alternatively FastAPI + SQLAlchemy + Alembic) - REST API framework
• Celery - distributed task queue
• PostgreSQL - relational database
• RabbitMQ - message broker
• Redis - in-memory data store

DevOps & Infrastructure

• uv (earlier Poetry + pyenv) - package management
• Docker + Docker Compose - containerization platform
• Pulumi - infrastructure as code
• Sentry - error tracking
• AWS - cloud platform
• Uvicorn, Gunicorn - application servers
• NGINX (or Traefik for containerized deployments) - load balancer and reverse proxy
• Let's Encrypt - SSL/TLS certificate authority
• GitHub Actions - CI/CD platform

Development & Quality Assurance Tools

• PyCharm - IDE
• AI coding assistants
  • Claude Code
  • Codex CLI
  • ChatGPT
  • GitHub Copilot
• Linear (alternatively - GitHub Projects) - project management
• pytest - testing framework
• diff-cover - test coverage tool
• pre-commit - git hook manager
• Ruff (earlier Flake8, Pylint) - linting and formatting
• GNU Make - command automation
• model-bakery - fixture factory
• ngrok - local tunneling service
• VCR.py - HTTP interaction recording
• Moto - AWS service mocking
• pdbpp - enhanced debugger
• colorlog - colored logging

Libraries

• django-split-settings - settings organization
• django-restql - GraphQL-like API queries
• django-dirtyfields - model field change tracking
• WhiteNoise - static file serving
• Anymail - email service integration
• python-jose - JWT library
• Jinja - template engine
• time-machine - time mocking for tests
• Pydantic - data validation
• tqdm - progress bars
• cachetools - caching utilities
• HTTPX (preferred over Requests) - HTTP client
• Tenacity - retry library
• Beautiful Soup - HTML/XML parser
• Liquid - template engine

General Software Development Principles

• Respect best practices
• Respect industry standards
• Respect backward compatibility
• Respect forward compatibility
• Write code to be read by humans (and AI), not just to be executed
• Continuous refactoring
• Continuous Learning
• Use AI for coding as much as possible
• Review and refactor AI-generated code
• The Less code (lines) the better
• DRY (Don't Repeat Yourself)
• KISS (Keep It Simple, Stupid)
• YAGNI (You Aren't Gonna Need It)
• Principle of Least Astonishment
• Code reuse
• Encapsulation
• Have a reason behind every decision
• Respect business needs over technical preferences
• Automate repetitive tasks
• Unit testing
• Consistency
• and more...

General Coding Rules

• AI Augmented Workflow
• Start new project with latest dependencies if possible
• Declare and assign variable values as close to the usages as possible
• Do not put a value to a variable if it is then read just once
• Do not implement logic that depends on environment name the code is running in
• Do not hard code a list of environments
• Avoid dead code
• Self code review and AI aided code review before submitting a pull request
• Prefer REST API over RPC API, otherwise use GraphQL API
• Use comments and docstrings reasonably
• Use asserts to describe expected invariants and assumptions in the code
• Use TODO or FIX comments to describe technical debt and convey intended imperfections to other developers
• Implement database schema changes and related business logic in the same pull request
• Use consistent data attributes naming
• Assign or declare variables as close to their usage as possible
• Use ServiceNameClient class wrapper even with official SDKs and service clients
• and more...

Python Specific Rules

• Follow PEP 8 coding style guide, read more here
• Use Python logging framework instead of print statements for logging
• In except clause always prefer .exception() over .error() logger method, unless you have a good reason for otherwise
• Use keyword arguments for arguments with default values. Call def f(arg='default'): with f(arg='value') instead of f('value').
• and more...