Interview Agent

An intelligent interview simulation system powered by LLMs that automatically generates relevant questions based on a candidate's CV, conducts interviews, and handles clarifications when needed.

Overview

This project implements an automated interview agent using LangChain and LangGraph. The system analyzes a candidate's CV to extract keywords, generates tailored interview questions, and conducts a natural conversation flow with clarification capabilities.

Features

• Keyword Extraction: Automatically identifies key skills and experiences from CV text
• Dynamic Question Generation: Creates relevant technical questions based on extracted keywords
• Interactive Interview Flow: Simulates a real interview experience with follow-up questions
• Clarification Handling: Intelligently responds when candidates ask for clarification
• Conversation Management: Maintains context throughout the interview process

Architecture

The system is built as a state machine using LangGraph with the following components:

• State Management: Tracks CV data, questions, answers, and conversation history
• Interviewer Agent: Handles keyword extraction, question generation, and conversation flow
• Decision Logic: Determines whether to continue asking questions, provide clarification, or end the interview

A visualization of the graph workflow is generated as graph.png when running the application.

Installation

1. Clone the repository
2. Create a virtual environment:

   python -m venv venv
   source venv/bin/activate  # On Windows: venv\Scripts\activate

3. Install dependencies:

   pip install -r requirements.txt

4. Set up your environment variables in a .env file

Usage

1. Ensure you have Ollama installed and the required LLM model (llama3.2:1b) available
2. Create prompt files in the prompts/ directory:
   • clarification_node.txt
   • decider_node.txt
   • fetch_keyword.txt
   • generate_question.txt
3. Run the application:

   python main.py

Customization

You can customize the interview agent by:

1. Modifying the CV data in main.py
2. Adjusting the prompt templates in the prompts/ directory
3. Changing the LLM model in the Interviewer initialization

Requirements

• Python 3.8+
• LangChain and LangGraph libraries
• Ollama for local LLM inference
• Additional dependencies listed in requirements.txt

License

This project is open source and available under the MIT License.