🤖 AI CV Customizer (Multi-Agent System)

This project implements an autonomous multi-agent system using CrewAI and Groq (powered by the Llama 3.1 8B model). Its main objective is to automate the process of tailoring a base curriculum vitae (CV) to perfectly align with the requirements and keywords (ATS filters) of a specific job opening.

Unlike a traditional single-block prompt, this software divides the cognitive load among specialized agents that simulate a professional assembly line.

---

🏗️ System Architecture and Flow

The system operates sequentially (Process.sequential), where the output of the first agent automatically becomes the input context for the second:

1. ATS Filter and Analysis: The Recruiter agent dissects the job description to identify what the company is truly looking for.
2. Writing and Optimization: The Writer agent receives the recruiter's findings and rewrites the key sections of the original CV, aligning the candidate's achievements with the job opening without inventing any information.

---

🌟 Key Features

• Multi-Agent Architecture: Segmented workflow for higher accuracy and a lower hallucination rate.
• Zero Cost (Free Tier): Natively integrated with the Groq API using the ultra-fast llama-3.1-8b-instant model.
• Compatibility Bypass: Configured using the OpenAI protocol redirected to Groq to avoid Pydantic data validation errors.
• Fault-Proof Shielding (Monkey Patch): Includes an interceptor function that cleans up cache properties (cache_breakpoint) injected by CrewAI before sending the request, preventing crashes due to incompatibility with the Groq backend.
• Structured Output: Generates the final result directly in Markdown format, ideal for subsequent conversion to PDF or HTML.

---

🛠️ Requirements and Installation

This project uses uv as a high-speed Python package manager.

1. Clone the repository and set up the environment

# Enter the project folder
cd agente-cv-crew

# Create and activate the virtual environment
uv venv
.venv\Scripts\activate  # On Windows (PowerShell)
source .venv/bin/activate  # On Linux/macOS

2. Install dependencies

uv pip install crewai litellm langchain-groq

---

🚀 Usage and Configuration

1. Open the src/app.py file.
2. Insert your free Groq API Key in the corresponding variable:

CLAVE_GROQ = "gsk_your_key_here"

3. Run the script from your terminal:

python src/app.py

---

📂 Main Code Structure (app.py)

• Section 0 (Monkey Patch): Intercepts litellm.completion to remove cache parameters that cause BadRequestError on Groq.
• Section 1 (LLM Setup): Initializes the LLM class pointing to api.groq.com.
• Section 2 (Agents): Defines the roles, goals, and backstories of the Recruiter and the Writer.
• Section 3 (Inputs): Variables where the user's original CV and the chosen job description are stored.
• Section 4 (Tasks): Defines the deliverables (expected_output) for each phase of the flow.
• Section 5 (Orchestration): Instantiates the Crew, bringing together the agents and tasks sequentially.