Environment setup
Welcome! I'm Param. In this course we are going to take a naive one-prompt CV analyzer and rebuild it as a real LangGraph workflow. By the end you will have specialized agents, typed state, conditional routing, and parallel execution all running on your laptop.
The project is a FastAPI service that takes a CV and a job description, then runs a graph of specialists. One agent extracts structure, one analyzes the job, one finds strengths, one finds weaknesses, one suggests improvements, one scores the fit, and one checks ATS compatibility. Every lesson is about a real decision you made while shaping that graph.
# Clone the workshop repository
git clone https://github.com/learnwithparam/multi-agent-langgraph.git
cd multi-agent-langgraph
# Set up the virtual environment and install dependencies
make devThe repo uses uv for fast installs. One command sets up the environment and starts the FastAPI server.
# Copy the example and pick any supported provider
cp .env.example .env
# Example with OpenRouter (free tier works)
LLM_PROVIDER=openrouter
OPENROUTER_API_KEY=your_key_here
OPENROUTER_MODEL=google/gemma-3-12b-itThe provider layer abstracts OpenRouter, Gemini, Fireworks, and OpenAI. Swap providers by changing environment variables. Your agent code never knows the difference.
Validation checklist: Environment setup checklist
Loading practice…