Skip to content

Installation

Warning

Arkalos has been released publicly as 0.1 (Beta 1), where the code may change or break without notice. Not all documentation and modules are complete. Use it only for experimentation, study, and practice, not for production use. Follow us and join the community to stay updated.

Requirements

  • Git
  • Python > 3.13
  • Python uv package manager > 0.5.29
  • Ollama - for downloading, running, and building LLMs locally.
  • IDE - like VS Code
  • Just basic coding skills. You don't have to know math, stats, ML/AI to use core Arkalos features.

For Windows Users

If you're on Windows, we recommend using Windows Subsystem for Linux (WSL).

Check out Microsoft's installation guide.

Git

You likely have Git installed already.

If not, follow GitHub's Installing Git guide.

To verify installation:

git --version

UV Package Manager

uv is a modern, super-fast Python package and project manager.

If you've used NodeJS (npm) or PHP (Composer), it's similar.

  • pyproject.toml is like package.json.
  • Use commands like uv add <package> or uv sync to manage dependencies.

To install uv:

curl -LsSf https://astral.sh/uv/install.sh | sh

powershell -ExecutionPolicy ByPass -c "irm https://astral.sh/uv/install.ps1 | iex"

For more details, see the uv documentation.

To verify installation:

uv -V

Python

Download and install the latest Python version from the official site.

Alternatively, use uv to install Python:

uv python install

To verify installation:

python -V

Installing Ollama and Open-Source AI Models

Ollama is a sleek, user-friendly app, CLI, framework, and repository for managing open-source LLMs locally.

Download and install Ollama.

To verify installation:

ollama -v

For more details, CLI commands, and RAM requirements, visit the Ollama GitHub repository.

Once installed, open a new terminal and start the Ollama server:

ollama serve

Now, you can download various open-source LLMs like DeepSeek, Llama, Mistral, and Gemma.

Open a new terminal and download your preferred model. Smaller models (~5GB) will take some time.

You'll need at least 8GB of RAM for 7B models.

Arkalos includes a basic AI Agent — a terminal chatbot to interact with your data warehouse.

We recommend downloading the qwen-2.5-coder model (7B by default) while continuing with this guide:

ollama pull qwen2.5-coder

We suggest using VS Code as your IDE because:

  • It supports Jupyter Notebooks and other extensions.
  • It includes a copilot.
  • For Windows and WSL users, it simplifies working inside WSL.

Download and install VS Code.

Install Extensions:

From the Extensions side tab, search for and install:

Official Microsoft Extensions:

  • Python
  • Python Debugger
  • Pylance
  • Mypy Type Checker
  • isort
  • Jupyter (and extension packs)
  • WSL (for Windows + WSL users)

Other Recommended Extensions:

  • Project Manager (by Alessandro Fragnani)
  • DotENV (by mikestead)
  • SQLite Viewer (by Florian Klampfer)
  • Excel Viewer (by MESCIUS)
  • Material Icon Theme (for a nicer look) (by Philipp Kief)

Create a New Arkalos Project

You're all set! Now, you can create a new Arkalos project.