My macbook setup for AI and Data Science development in 2026. A lot has changed in AI development since I set up my last macbook pro, and all of the tools below are based on how my workflow as evolved. As such this is highly personal to my workflow and yours might be different.

The Mac does not have a great official package and Homebrew is the first thing I have installed on a mac

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

A lot of packages have additional dependencies so update xcode before anything else

xcode-select --install

Install and active mise - the package manager for your dev tools

brew install mise
echo 'eval "$(mise activate zsh)"' >> ~/.zshrc
source ~/.zshrc

In 2025, I switched from iTerm to Ghostty. It's fast, lightweight and has everything I need without the distractions. You can install it via the universal binary @ - https://ghostty.org/download or a community maintained package via homebrew

Optional, I like it for my CLI use

git clone https://github.com/powerline/fonts.git --depth=1
cd fonts
./install.sh
cd ..
rm -rf fonts

I like the zsh shell and "oh-my-zsh" adds punch to it the base zsh shell

sh -c "$(curl -fsSL https://raw.githubusercontent.com/ohmyzsh/ohmyzsh/master/tools/install.sh)"

Lookup the .zshrc in the .dotfiles for additional custom settings

While I don't actively use iterm, but I have it installed via the official binary and with the following default profile settings

Profiles -> Text -> Font `Melso LG S DZ Powerline` Regular 14

Despite tabs and terminal splits, I still find myself using tmux, specially useful for running things in the background

brew install tmux

Command line video processing including things such as increasing video speed, adding images to videos and merging videos

brew install ffmpeg

Command line image processing, specially useful for image processing for ML training

brew install imagemagick

This is my default browser for most use cases and love the container support. Install via the official binary via - https://www.firefox.com/en-US/download/ or via homebrew

brew install --cask firefox

For the times you still need chrome, you can get via the official site @ https://www.google.com/chrome/ or via homebrew

brew install --cask google-chrome

This can get bloated very quickly and I am trying to stick with the bare minimum here. These extension are compatible across the both the browsers

Lastpass - https://www.lastpass.com/

ublock Origin lite - https://ublockorigin.com/

Privacy Badger - https://privacybadger.org/

I love docker, but docker on Mac is not the most efficient. Orbstack is a faster and a much more efficient way to run containers on Mac with support for Docker commands. Easy install via Homebrew

brew install orbstack
orb start
docker run --rm hello-world

All optional, but these are make quality of life improvements for me on the Mac

Something to backup and sync, by choice is dropbox but you can use what you like. Install via the the official dropbox site

Completely optional and personal, but Obsidian is the one place I do everything. From notes to task/project management, wireframe diagrams, knowledge base and even my shopping list. Everything that is important, eventually goes in here. It's free , open source and does not have any proprietary format - just uses markdown. Install via here - https://obsidian.md/

Obsidian is great out of the box, but these make my workflow bette

• Boards
• Mindmap
• Natural Language Dates
• Excalidraw

I have been using Alfred over a decade now, it makes spotlight much better with handy shortcuts and power packs. Install via the official site here - https://www.alfredapp.com/

Similar spotlight replacement with additional features and AI. I am still trying this one out - https://www.raycast.com/

Communities , messaging and 1:1 - https://slack.com/ or http://discord.com/

PyTorch’s MPS backend lets you run on the Apple GPU by moving tensors/models to the mps device

Install in your project via uv

uv add torch torchvision torchaudio

Minimal MPS test:

uv run python - <<'PY'
import torch
print("mps available:", torch.backends.mps.is_available())
x = torch.randn(1024, 1024, device="mps")
print(x @ x.T)
PY

Apple’s native ML framework

Install via homebrew

brew install ollama
ollama --version

Run a model locally

ollama pull llama3.2
ollama run llama3.2

Sometimes you need a UI. Open WebUI supports Ollama and OpenAI-compatible APIs (so you can point it at local or remote model servers).

docker pull ghcr.io/open-webui/open-webui:main
docker run -d -p 3000:8080 \
  -v open-webui:/app/backend/data \
  --name open-webui \
  ghcr.io/open-webui/open-webui:main

Ollama does most of this, but if you need more granular GPU offloading

brew install llama.cpp

Still using VS Code as the primary editor , install via the official site - https://code.visualstudio.com/download or Homebrew

brew install --cask visual-studio-code

• Python (Microsoft)
• Jupyter (Microsoft)
• Remote SSH
• AWS (Amazon)
• MCP

Kiro is my agentic AI IDE of choice with a generous free tier and the ability to switch between models. You can pick yours. I install mine from the offciall site here - https://kiro.dev/

• Postman
• AWS
• MCP

In 2026, Claude code is still the leader IMO for CLI based agentic coding. There is homebrew option, but I go for the native install

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

Eventually you will run out of local resources and will need to scale to the cloud, and again in 2026 the leader here is still AWS with a very generous free tier. Install the AWS CLI and setup a default profile

curl "https://awscli.amazonaws.com/AWSCLIV2.pkg" -o "AWSCLIV2.pkg
sudo installer -pkg AWSCLIV2.pkg -target /

Should be installed by default on the Mac, recommend to use the same

The default git on Mac works well with GitHub, the github cli adds the native github experience to your CLI

brew install gh

Another switch for 2025, I moved from PIP and Conda to uv for managing python environments and packages. I start by installing uv first and then use it to install and manage python versions and packages without messing up any of the global installs.

brew install uv
uv python install

A typical project workflow looks something like this

mkdir myproj && cd myproj
uv init #will setup a venv environment
uv add httpx #add packages similar to pip

As a bonus, you can use uv to install other tools like ruff without polluting the global pip

uv tool install ruff@latest

Love them or hate them, you will need to work with these. My way of managing these is via the individual project environments and uv An example workflow below

mkdir myproj && cd myproj
uv init
uv run --with jupyter jupyter lab

If you like reactive notebooks and “git-friendly” workflows, I recently discovered marimo

uv tool install marimo
marimo tutorial

A fast lightweight and good database alternative for data science.

curl https://install.duckdb.org | sh

A universal SQL client for the Mac - https://tableplus.com/download/

Local vector DB. Add via uv inside your projects

uv add chromadb

NoSQL cache DB

brew install redis

A knowledge base and graph DB. Install via homebrew

brew install neo4j