Prerequisites & Setup

[jrbrowning]

Prerequisites & Setup

This course is validated on macOS (developed on a MacBook Pro M-series with 36GB RAM).

Windows/Linux users are welcome, but may need to adapt certain commands. Please post any useful tips here.

[jrbrowning]

Required Software

• Docker Desktop (26GB Memory Limit)
• Python 3.12+ (for this repo)
• Node.js 20+ (for frontend repo. not needed until week 3)
• Ollama (installation guide below)

All models are defined inside the central .env

TRADITIONAL_MODEL=llama3.1:8b
TRADITIONAL_MODEL_ALT=qwen2.5-coder:3b
REASONING_MODEL=deepseek-r1:8b
REASONING_MODEL_ALT=qwen3:1.7b

Full Development Setup

1. Pull the repository
   git clone git@github.com:jrbrowning/ollama-tools-api.git
2. Python environment: This course uses Poetry for dependency management

cd ollama-tools-api
source bin/setup_poetry_env.sh

poetry --version
# Poetry (version 2.1.4)  (August 2025)

3. Docker: Must be installed and running

which docker
# Should output: //usr/local/bin/docker

docker info  # Should return system information
# Example:    Version:  28.3.2 (August 2025)

If this fails, you'll need to install Docker Desktop and possibly symlink the CLI. Check Docker's documentation for your system.

4. Ollama On macOS, Docker containers (as of August 2025) do not have access to GPU resources, only CPU.

Best Practice

• Adjust docker-compose files as needed for your workflow based on memory available (see below)

NOTE: As of August 2025, the Homebrew native installation is pinned to an outdated version of Ollama. As a result, some manual steps are required to install the latest version in the location Ollama expects. This is only needed once.

4.1. Install Ollama from the official source:
Download from https://ollama.com/download and install.

4.2. Open Ollama.app once to initialize.

4.3. Manually softlink Ollama binary:.
Note: This command creates a symbolic link, which can be easily removed later, and points the application to the expected location as if it were installed officially via Homebrew. Only use sudo when necessary, as it carries elevated permissions. For this reason, I have wrapped it to avoid accidental cut-and-paste. Remove () to use.

 (sudo) ln -sf /Applications/Ollama.app/Contents/Resources/ollama /opt/homebrew/bin/ollama

4.4. Verify installation:

 which ollama
 # Should output: /opt/homebrew/bin/ollama

 ollama --version
 # Example: ollama version is 0.11.4 (August 2025)

)

Docker Setup

 docker-compose build; docker-compose up

(This will take several minutes as you are pulling all 4 models)

You should see 4 containers.

Container 5 is webUI to interface with the models using a standard chat interface.

 http://localhost:8080/

NOTE: Ollama is local, but still requires a login. You will need to do this once

1. Click on Sign up
   ( Don't have an account? Sign up )
2. You can create anything here for your login. You don't need to use a real email or name.

You are now good to go!

---

Support Expectations

Stuck? Post in Q&A with your specific error and what you've tried. The community and I will help when we can, but environment setup issues are great learning opportunities - the debugging skills you develop here will serve you throughout your career.

The self-paced option will always be here when you're ready.