← For Students

Set up your environment

Work through these once and you're set for everything else. Pick your operating system in each block. The Copy button grabs just those commands. If a command fails, paste the exact error into a search or an AI assistant; don't spin on it.

01 A package manager

A package manager installs everything else with one command. macOS uses Homebrew; most Linux distros ship one already; Windows 10/11 include winget.

/bin/bash -c "$(curl -fsSL https://raw.githubusercontent.com/Homebrew/install/HEAD/install.sh)"
brew --version

# Debian/Ubuntu already has apt; just refresh it
sudo apt update

# winget is built into Windows 10/11 — run in PowerShell
winget --version

02 Core tools: Git, uv, Node, VS Code

Git for version control (make a GitHub account too), uv for Python + dependencies, Node for some CLIs, and VS Code as your editor.

brew install git uv node
brew install --cask visual-studio-code
git --version && uv --version

sudo apt install -y git nodejs npm
curl -LsSf https://astral.sh/uv/install.sh | sh   # uv
# VS Code: install from code.visualstudio.com
git --version && uv --version

winget install Git.Git OpenJS.NodeJS Microsoft.VisualStudioCode
powershell -c "irm https://astral.sh/uv/install.ps1 | iex"   # uv
git --version; uv --version

03 Ollama: run open models locally

Ollama runs open-weight LLMs entirely on your machine: no API key, no per-token cost, and nothing leaves your computer.

# Download the app from https://ollama.com/download, then:
ollama pull gemma4      # or any model from ollama.com/library
ollama run gemma4       # opens a chat; type /bye to exit
ollama list

curl -fsSL https://ollama.com/install.sh | sh
ollama pull gemma4      # or any model from ollama.com/library
ollama run gemma4       # opens a chat; type /bye to exit
ollama list

# Download the app from https://ollama.com/download, or:
winget install Ollama.Ollama
ollama pull gemma4      # or any model from ollama.com/library
ollama run gemma4       # /bye to exit
ollama list

04 Claude Code: an AI coding assistant

Reads a codebase and helps you write, run, and debug. Invaluable when you're new to a large repo. Needs a paid Claude plan or API access.

curl -fsSL https://claude.ai/install.sh | bash
claude --version && claude doctor

curl -fsSL https://claude.ai/install.sh | bash
claude --version && claude doctor

# run in PowerShell
irm https://claude.ai/install.ps1 | iex
claude --version; claude doctor

Any OS with Node 18+ can instead run npm install -g @anthropic-ai/claude-code. If claude isn't found after install, open a new terminal so your PATH refreshes. Setup docs: code.claude.com/docs/en/setup.

05 Cloud GPUs: when your laptop isn't enough

No install needed. Start on Google Colab for zero-setup GPU notebooks, and graduate to a persistent environment like Lightning AI Studio for sustained project work. Keep data and checkpoints out of git, and save them before a session disconnects.