Dev Environment Setup

Before you write your first agent, you need four things working: Python, a code editor, a virtual environment, and Ollama running locally. This module walks you through all four — step by step, with checkpoints.

1 · Install Python 3.11+

All code in this course requires Python 3.11 or later. Python 3.12 is recommended — it has better error messages and faster startup.

python --version       # Should print: Python 3.11.x or 3.12.x
python -m pip --version  # Should print: pip 24.x from ...

# Install Python 3.12
brew install python@3.12

# Verify (may need to restart terminal first)
python3 --version     # Python 3.12.x
python3 -m pip --version

python3 --version     # Check current version

# Install 3.12 if needed
sudo apt update && sudo apt install python3.12 python3.12-venv python3-pip

# Verify
python3.12 --version

2 · Choose Your IDE

For this course you'll use either VS Code or PyCharm Community (both free). Both work equally well — pick the one you prefer or already have installed.

The next two sections cover setup for each. Skip the one you're not using.

3 · VS Code Setup

mkdir ai-agents-course
cd ai-agents-course

4 · PyCharm Setup

5 · Create a Virtual Environment

(Skip this if you used PyCharm's built-in venv creation in Step 2 above.)

Run these commands in your project folder. The venv folder created is about 20MB and contains an isolated Python installation.

# Navigate to your project folder first
cd ai-agents-course

# Create the virtual environment
python -m venv venv

# Activate it (you must do this every new terminal session)
.\venv\Scripts\Activate.ps1

# Verify — prompt should show (venv)
python --version

# Navigate to your project folder first
cd ai-agents-course

# Create the virtual environment
python3 -m venv venv

# Activate it
source venv/bin/activate

# Verify — prompt should show (venv)
python --version   # Note: just "python" now works (not python3)

6 · Install Dependencies

With your virtual environment active, install the packages needed for this course. The openai package is the primary SDK — it connects to Ollama, Groq, Together AI, and any other OpenAI-compatible server.

# Core package — works with Ollama, Groq, Together AI
pip install openai

# For RAG modules (M09): local embeddings + vector store
pip install sentence-transformers chromadb

# For environment variable management (recommended)
pip install python-dotenv

# Verify everything installed
pip list | grep -E "openai|sentence|chroma|dotenv"

Expected output from pip list:

chromadb          0.5.x
openai            1.x.x
python-dotenv     1.x.x
sentence-transformers  3.x.x

Save your dependencies

Run this after installing to create a requirements.txt — anyone else can reproduce your environment exactly with pip install -r requirements.txt.

pip freeze > requirements.txt

7 · Install Ollama & Pull Mistral-7B

Ollama is the local inference server. Once installed it runs as a background service, exposing a REST API on http://localhost:11434. The Python openai SDK talks to this same URL — Ollama speaks the OpenAI Chat Completions protocol.

# Pull Mistral-7B (4.1 GB download — do this on Wi-Fi)
ollama pull mistral

# Verify the model is available
ollama list

# Test a quick response
ollama run mistral "Say hello in one sentence."

# Install via Homebrew (recommended)
brew install ollama

# OR download the macOS app from ollama.ai

# Start the server (runs in background)
ollama serve &

# Pull Mistral-7B
ollama pull mistral

# Verify
ollama list

# Install with the official script
curl -fsSL https://ollama.ai/install.sh | sh

# Start the server
ollama serve &

# Pull Mistral-7B (CUDA GPU detected automatically if available)
ollama pull mistral

# Verify
ollama list
curl http://localhost:11434/api/tags   # Should return JSON with model info

8 · Verify Everything Works

Create a new file called hello_mistral.py in your project folder and paste the code below. This is your "all systems go" test.

"""
hello_mistral.py — verify your dev environment is working
Run: python hello_mistral.py
Expected: a one-sentence response from Mistral-7B running locally
"""
from openai import OpenAI

# Connect to Ollama running on localhost
# api_key="ollama" is a placeholder — Ollama doesn't need a real key
client = OpenAI(
    base_url="http://localhost:11434/v1",
    api_key="ollama"
)

print("Connecting to local Mistral-7B via Ollama...")
print("-" * 50)

try:
    response = client.chat.completions.create(
        model="mistral",
        messages=[
            {"role": "system", "content": "You are a helpful assistant. Be concise."},
            {"role": "user", "content": "In exactly one sentence, what is a large language model?"}
        ]
    )
    text = response.choices[0].message.content
    print(f"Mistral says: {text}")
    print("-" * 50)
    print(f"Tokens used — input: {response.usage.prompt_tokens}, output: {response.usage.completion_tokens}")
    print("\n✅ Setup complete! Your environment is ready for all 12 modules.")

except Exception as e:
    print(f"❌ Error: {e}")
    print("\nTroubleshooting:")
    print("  1. Is Ollama running? Run: ollama serve")
    print("  2. Is mistral pulled? Run: ollama pull mistral")
    print("  3. Is openai installed? Run: pip install openai")
    print("  4. Is venv active? Check for (venv) in your prompt")

python hello_mistral.py

9 · Recommended Project Structure

Organize your work like this from the start. Each module gets its own folder inside labs/. This keeps your files tidy and makes it easy to compare your solution to the reference code.

Create the folder structure now

New-Item -ItemType Directory -Force labs/m01-llm-mental-model,labs/m03-prompts,labs/m05-function-calling,labs/m08-conversation,labs/m09-rag,labs/m12-react,labs/m13-planning,labs/m14-multi-agent,labs/m16-guardrails-in,labs/m17-guardrails-out

# Create .gitignore
@"
venv/
.env
__pycache__/
*.pyc
.DS_Store
"@ | Out-File -Encoding utf8 .gitignore

# Create .env template
@"
# Groq (optional — free tier at console.groq.com, faster than local Ollama)
# GROQ_API_KEY=your_key_here

# Together AI (optional)
# TOGETHER_API_KEY=your_key_here
"@ | Out-File -Encoding utf8 .env

mkdir -p labs/{m01-llm-mental-model,m03-prompts,m05-function-calling,m08-conversation,m09-rag,m12-react,m13-planning,m14-multi-agent,m16-guardrails-in,m17-guardrails-out}

# Create .gitignore
cat > .gitignore << 'EOF'
venv/
.env
__pycache__/
*.pyc
.DS_Store
EOF

# Create .env template
cat > .env << 'EOF'
# Groq (optional — free tier at console.groq.com)
# GROQ_API_KEY=your_key_here
EOF

Knowledge Check

Five quick questions to confirm your setup understanding before starting the labs.