Employee Recruiter Agent 🤖

An enterprise-grade, AI-powered recruitment pipeline that automates the entire candidate lifecycle: from resume ingestion and parsing to intelligent screening and automated interview scheduling.

---

🚀 Overview

The Employee Recruiter Agent is a sophisticated autonomous system built on LangGraph. It orchestrates a multi-step workflow designed to reduce the overhead of technical recruiting. By leveraging LLMs (Groq/Gemini) and asynchronous task processing, it ensures high throughput and stateful execution.

Key Capabilities

• 📄 Intelligent Parsing: Automatically extracts structured JSON data from PDF resumes using specialized LLM prompts.
• 🎯 Advanced Screening: Scores candidates against job descriptions with objective justification.
• 🧠 Autonomous Decisioning: Uses configurable thresholds to decide between scheduling interviews or sending rejections.
• ⏱️ Asynchronous Processing: Handles heavy LLM tasks in the background using Redis and RQ.
• 💾 State Persistence: Maintains workflow state across restarts using MongoDB.

---

🏗️ Architecture

The system follows a directed acyclic graph (DAG) structure managed by LangGraph:

graph TD
    START((Start)) --> Load[Load Resume]
    Load --> Parse[Parse Resume]
    Parse --> Screen[Screen Candidate]
    Screen --> Decide{Decision}
    
    Decide -->|Score >= Threshold| Schedule[Schedule Interview]
    Decide -->|Score < Threshold| Reject[Send Rejection]
    
    Schedule --> Invite[Send Invite]
    Invite --> END((End))
    Reject --> END

Loading

Components

• API Layer: FastAPI handles incoming requests and provides job status monitoring.
• Worker Layer: RQ (Redis Queue) processes the LangGraph workflows asynchronously.
• Persistence Layer: MongoDB stores the state of each recruitment "thread".

---

🛠️ Tech Stack

• Core Framework: LangGraph, LangChain Community
• API Framework: FastAPI, Uvicorn
• Task Queue: Redis, RQ (Python-RQ)
• Database: MongoDB (State Checkpointing)
• LLM Providers: Groq (Default), Google Gemini (Optional)
• Document Processing: PDFMiner

---

⚙️ Getting Started

Prerequisites

Ensure you have the following installed and running:

• Python 3.10+
• Redis Server (brew install redis or docker run -p 6379:6379 redis)
• MongoDB (Local or Atlas)

Installation

1. Clone the repository:

   git clone https://github.com/yourusername/EmployeeRecuiterAgent.git
   cd EmployeeRecuiterAgent

2. Initialize Environment: Using uv (recommended) or pip:

   python3 -m venv .venv
   source .venv/bin/activate
   pip install -r requirements.txt

3. Configure Environment Variables: Create a .env file from the example:

   cp .env.example .env

   Required variables:

   GROQ_API_KEY=your_groq_api_key
   MONGODB_URI=mongodb://localhost:27017/recruiter_agent
   REDIS_URL=redis://localhost:6379

Running the System

The project includes a Makefile and an orchestration script for professional management:

# Option 1: Start everything (API + Worker)
make run

# Option 2: Individually manage services
make api       # Start FastAPI only
make worker    # Start RQ Worker only

Alternatively, use the startup script directly:

chmod +x start.sh
./start.sh

---

🛠️ Development & Standards

This project adheres to professional development standards:

• Logging: Integrated Python logging throughout for better observability.
• Robust Env: Specialized shell parsing in start.sh to handle complex .env values.
• Clean Code: Type hinting and structured error handling in all core utilities.
• Workflow: Automated service management with automated cleanup on exit.

Management Commands

Command	Description
make install	Install all dependencies
make test	Run the pytest suite
make clean	Remove logs and temporary Python files

---

1. Synchronous Execution

Execute the workflow and wait for the final result.

Endpoint: POST /execute_workflow

Payload:

{
  "candidate_info": {
    "name": "Jane Doe",
    "email": "jane@example.com",
    "phone": "+123456789",
    "resume_url": "https://example.com/resume.pdf"
  },
  "thread_id": "unique-session-id-001",
  "job_spec": {
    "title": "Senior Software Engineer",
    "description": "Expert in Python, FastAPI, and AI agents.",
    "required_skills": ["Python", "LangChain"]
  }
}

2. Asynchronous Execution (Recommended)

Enqueue the workflow and get a job_id.

Endpoint: POST /rq/workflow

Response:

{
  "status": "ok",
  "job_id": "550e8400-e29b-41d4-a716-446655440000"
}

3. Check Job Status

Monitor the progress of a background job.

Endpoint: GET /rq?job_id=<job_id>

---

🛠️ Development

Customizing Workflow

The graph logic is defined in queues/graph.py. You can adjust the threshold in the decide node or add new nodes for additional steps (e.g., background checks, coding tests).

Testing

Run the test suite:

pytest tests/

---

📄 License

This project is licensed under the MIT License - see the LICENSE file for details.

---

Built with ❤️ for the recruitment community.