🧑‍💼 Candidate Screening Workflow with LangGraph + LangSmith Tracing

This project is an AI-powered candidate screening system built with LangGraph, LangChain, LangSmith, and OpenAI.
It automatically evaluates job applications, categorizes candidates by experience level, checks for a skill match, and then decides whether to:

• ✅ Shortlist for HR interview
• 📈 Escalate to a recruiter
• ❌ Send a rejection email

---

✨ Features

• Integrated LangSmith tracing – monitor, debug, and visualize workflow execution
• State machine workflow with LangGraph
• Experience level classification (Entry-level | Mid-level | Senior-level)
• Skillset assessment (Match | No Match)
• Dynamic routing (HR → Recruiter → Rejection)
• Extensible design – easily adapt prompts for other job roles

---

🛠 Tech Stack

• LangSmith – tracing & monitoring
• Python 3.10+
• LangGraph
• LangChain
• OpenAI / ChatOpenAI
• python-dotenv

---

📂 Folder Structure

.
├── main.py   # Main workflow
├── .env.example              # Stores API keys (ignored in git)
├── requirements.txt         # Dependencies
└── README.md                # Project documentation

---

⚙️ Setup Instructions

1. Clone this repo

   git clone https://github.com/<your-username>/candidate-screening.git
   cd candidate-screening

2. Create virtual environment

   python -m venv venv
   source venv/bin/activate  # Mac/Linux
   venv\Scripts\activate     # Windows

3. Install dependencies

   pip install -r requirements.txt

4. Add API key
   Create a .env file in the project root:

   OPENAI_API_KEY=your_api_key_here
   

---

▶️ Usage

Run the workflow with a sample candidate application:

python main.py

Example Output

Categorizing the experience level of the candidate
Assessing the skillset of candidate :
Skill Match : No Match
Sending Rejection Email


Computed Results :
Application: I have 10 years of experience in software engineering with expertise in C#
Experience Level: Senior-level
Skill Match: No Match
Response: Candidate has been escalated to the recruiter.

---

🔍 LangSmith Integration

LangSmith tracing is enabled in this project to:

• Monitor each step of the LangGraph workflow
• Debug model inputs/outputs
• Visualize execution traces for better observability

---

🔮 Future Improvements

• ✅ Structured output parsing (guarantee clean values)
• ✅ Configurable job roles instead of hardcoding "Python Developer"
• ✅ Error handling & validation
• ✅ REST API / frontend integration

---

🤝 Contributing

Feel free to fork this repo and submit PRs with improvements 🚀

---

📜 License

MIT License © 2025 [Kartik Garg]