AI Interview Agent 🤖

An intelligent AI-powered interview system that analyzes candidate resumes against job descriptions and conducts personalized, interactive interviews using Google's Gemini AI and LangGraph workflow orchestration.

🚀 Features

📊 Intelligent Analysis

• Resume-JD Compatibility Analysis: Evaluates alignment between candidate background and job requirements
• Skills Gap Identification: Identifies missing skills and extra strengths
• Experience Matching: Analyzes relevant experience and domain expertise
• PDF Processing: Automatically extracts text from resume and job description PDFs

🎯 Dynamic Question Generation

• Technical Questions: Based on candidate's technical background and job requirements
• Behavioral Questions: STAR method-based questions focusing on soft skills and leadership
• Situational Questions: Scenario-based problem-solving challenges
• Context-Aware: Questions tailored to candidate's specific experience and role

💬 Interactive Interview Experience

• Conversational Interface: Natural, AI-driven interview conversations
• Follow-up Questions: Dynamic questioning based on candidate responses
• Real-time Evaluation: Continuous assessment during the interview
• Multi-modal Interface: Both web-based (Streamlit) and command-line options

📈 Comprehensive Evaluation

• Response Analysis: AI-powered evaluation of candidate answers
• Depth Assessment: Measures the depth and quality of responses
• Interview Summary: Detailed evaluation report with recommendations
• JSON Export: Structured data export for further analysis

🏗️ Architecture

Workflow Structure

📁 AI-Interview-Agent/
├── 🎯 main.py                      # Command-line entry point
├── 🌐 streamlit_app.py             # Web interface 
├── 🏃 run_streamlit.py             # Streamlit launcher script
├── 🧠 llm.py                       # Google Gemini AI integration
├── 🔄 graph.py                     # LangGraph workflow orchestration
├── 📋 requirements.txt             # Python dependencies
├── 📁 tools/                       # Core processing modules
│   ├── 🔍 analyzer.py              # Resume-JD compatibility analysis
│   ├── 📊 context_splitter.py      # Context categorization engine
│   ├── ❓ question_generator.py    # Interview question generation
│   └── 💬 interview_conversational_agent.py # Interactive interview conductor
├── 📁 resources/                   # Sample documents
│   ├── 📄 cv-ml.pdf               # Sample resume
│   └── 📄 JD-ml.pdf               # Sample job description
└── 📖 README.md                    # This documentation

Processing Pipeline

graph TD
    A[📄 PDF Resume] --> B[📋 To Markdown ]
    C[📄 PDF Job Description] --> B
    B --> D[🔍 Resume-JD Analysis]
    D --> E[📊 Context Splitting]
    E --> F[❓ Question Generation]
    F --> G[💬 Interactive Interview]
    G --> H[📈 Response Evaluation]
    H --> I[📊 Final Assessment]

Loading

⚙️ Installation

Prerequisites

• Python
• Google API key for Gemini AI

Setup Steps

1. Clone the repository

   git clone https://github.com/gopalkrishna2004/AI-Interview-Agent.git
   cd AI-Interview-Agent

2. Install dependencies

   pip install -r requirements.txt

3. Configure API Key

   Edit llm.py and replace the placeholder:

   os.environ["GOOGLE_API_KEY"] = "your-actual-api-key-here"

   Or set as environment variable:

   export GOOGLE_API_KEY="your-actual-api-key-here"

4. Verify installation

   python run_streamlit.py

🎮 Usage

Option 1: Web Interface (Recommended)

Launch the Streamlit web application:

python run_streamlit.py

Then open your browser to http://localhost:8501

Features:

• 📤 File Upload: Upload resume and job description PDFs
• 🎨 Modern UI: Clean, professional interface
• 📊 Visual Results: Interactive display of analysis and questions
• 💬 Live Chat: Real-time interview conversation
• 📱 Responsive Design: Works on desktop and mobile

Option 2: Command Line

For programmatic usage or testing:

python main.py

Note: Update the file paths in main.py:

resume_pdf_path = "path/to/your/resume.pdf"
jd_pdf_path = "path/to/your/job_description.pdf"

🔧 Core Components

1. LLM Integration (llm.py)

• Google Gemini 2.5 Flash: High-performance AI model integration
• Standardized Interface: Consistent AI interaction across all modules
• Error Handling: Robust API communication

2. Workflow Orchestration (graph.py)

• LangGraph Framework: State-based workflow management
• Two-Phase Processing:
  • Phase 1: Analysis → Context Split → Question Generation
  • Phase 2: Interactive Interview → Evaluation
• State Management: Comprehensive data flow between components

3. Resume-JD Analyzer (tools/analyzer.py)

# Key Analysis Areas:
✅ Matching Skills        # Skills alignment assessment
📈 Experience Alignment   # Background compatibility  
🔬 Research Fit          # Domain expertise evaluation
⚠️ Skill Gaps           # Missing requirements identification
⭐ Extra Strengths       # Additional candidate value

4. Context Splitter (tools/context_splitter.py)

Categorizes candidate background into three contexts:

• 🔧 Technical Context: Skills, projects, technical experience
• 👥 Behavioral Context: Leadership, teamwork, communication
• 🎯 Situational Context: Problem-solving, challenges, decision-making

5. Question Generator (tools/question_generator.py)

• Technical Questions: Role-specific technical assessments
• Behavioral Questions: STAR method-based soft skill evaluation
• Situational Questions: Scenario-based problem-solving
• Customizable Difficulty: Adjustable complexity levels

6. Interview Agent (tools/interview_conversational_agent.py)

• Natural Conversation Flow: Human-like interview interactions
• Dynamic Follow-ups: Context-aware follow-up questions
• Real-time Evaluation: Continuous candidate assessment
• Conversation Memory: Maintains context throughout interview

📊 Output & Results

Analysis Report

{
  "compatibility_score": 85,
  "matching_skills": ["Python", "Machine Learning", "Docker"],
  "missing_skills": ["Kubernetes", "AWS"],
  "experience_alignment": "Strong background in ML engineering",
  "recommendation": "Highly suitable candidate"
}

Generated Questions

• Technical: "Explain your experience with distributed systems and how you've handled scalability challenges."
• Behavioral: "Tell me about a time when you had to lead a cross-functional team through a difficult project."
• Situational: "If you discovered a critical bug in production right before a major release, how would you handle it?"

Interview Evaluation

{
  "overall_score": 4.2,
  "technical_competency": 4.5,
  "communication_skills": 4.0,
  "problem_solving": 4.3,
  "cultural_fit": 4.1,
  "recommendations": ["Strong technical skills", "Good communication"]
}

🛠️ Configuration

Customizing Question Difficulty

Edit tools/question_generator.py:

difficulty_level = "advanced"  # Options: basic, intermediate, advanced

Adjusting Interview Flow

Modify tools/interview_conversational_agent.py:

question_type_order = ["technical", "behavioral", "situational"]
max_follow_ups = 2
conversation_depth = 3

API Configuration

Update model settings in llm.py:

model_name = "gemini-2.5-flash"  # or "gemini-pro"
temperature = 0.7
max_tokens = 1000

📁 Dependencies

Core Frameworks

• streamlit - Web interface framework
• langchain - AI application framework
• langchain-google-genai - Google Gemini integration
• langgraph - Workflow orchestration
• langgraph-checkpoint - State management
• langgraph-prebuilt - Pre-built components

PDF Processing

• PyMuPDF - PDF text extraction
• pymupdf4llm - LLM-optimized PDF processing

Utilities

• tempfile2 - Temporary file handling
• typing - Type hints and annotations

🚀 Getting Started Example

1. Prepare your documents: Have a resume PDF and job description PDF ready
2. Start the application: python run_streamlit.py
3. Upload files: Use the web interface to upload your documents
4. Review analysis: Check the compatibility assessment
5. Start interview: Begin the interactive interview session
6. Get evaluation: Receive comprehensive feedback and recommendations

🤝 Contributing

1. Fork the repository
2. Create a feature branch: git checkout -b feature-name
3. Make your changes
4. Test thoroughly
5. Submit a pull request