Python for AI Starter Toolkit

This is the practical setup page: everything you need installed and understood before writing your first line of AI-related Python. It is for beginners who have decided to learn Python for data work or AI and want a working environment without wading through conflicting advice. The whole setup takes under an hour, and every piece here is free.

Install Python from python.org

Download the current stable release from the official site, python.org. During installation on Windows, check the box that adds Python to your PATH — missing that checkbox causes most beginner setup problems. On a Mac, the python.org installer works the same way. Afterward, open a terminal and run "python --version" (or "python3 --version" on Mac) to confirm it responds.

You will see other ways to get Python — Anaconda, system packages, the Microsoft Store. They all work, but the python.org installer is the simplest to reason about and the one most tutorials assume.

Use a virtual environment for every project

A virtual environment is a private copy of Python's package space for one project, so the libraries you install for project A cannot break project B. The habit is simple: in each new project folder, run "python -m venv .venv", then activate it (the activate script lives in .venv's Scripts folder on Windows, bin on Mac and Linux). Your terminal prompt shows the environment name when it is active. This feels like ceremony at first; it becomes automatic within a week and saves you from the dependency tangles that make beginners reinstall everything.

Install libraries with pip

pip is Python's package installer and comes with Python. With your virtual environment active, "pip install pandas" downloads and installs the pandas library and everything it depends on. That is the entire mental model: one command per library, always inside the environment.

Get an editor: VS Code

Visual Studio Code (code.visualstudio.com) is free, works on every platform, and is the most common editor in tutorials, so instructions will match what you see. Install it, then add the official Python extension from the extensions panel — it gives you error highlighting, autocompletion, and a run button. VS Code also opens Jupyter notebooks, the interactive file format most data science material uses, so you do not need a separate notebook tool.

The three starter libraries

Install these first and ignore the rest until a tutorial demands them.

• NumPy — fast math on arrays of numbers. You will rarely use it directly at first, but nearly everything else is built on it.
• pandas — tables. Loading CSV files, filtering rows, grouping, summarizing, and cleaning data. This is where you will live for your first months.
• scikit-learn — classical machine learning: training a model, splitting data for testing, and measuring accuracy, all with a consistent and beginner-friendly design. When you reach it, machine learning in plain English explains the concepts behind the code.

When you eventually work with language models, libraries from Hugging Face (huggingface.co) are the standard next layer — but that is a later chapter, not setup.

What to build first

Prove the toolkit works end to end: create a project folder, make and activate a virtual environment, install pandas, and write a script that loads a CSV file and prints the number of rows and the column names. Then extend it — compute an average, count values in a category, find the largest entry. Small and finished beats ambitious and abandoned. Our Python starter project gives you a ready-made structure, the Python for AI tutorial teaches the language itself, and the Python for AI beginner roadmap sequences the months after that.