AI Interview Analysis Platform

A comprehensive AI-powered platform for interview analysis with skill assessment, Q&A grading, and detailed insights. Also supports general video transcription and formatting.

Features

🎯 Comprehensive Interview Analysis

• Skill Assessment: Evaluate specific skills with confidence scores and evidence
• Q&A Extraction: Automatic extraction and grading of interview questions and answers
• Performance Insights: Detailed analysis including communication clarity, technical depth, and hiring recommendations
• Structured AI Responses: Uses OpenAI's structured output for consistent, reliable analysis

📝 Transcript Extraction

• Multi-format Support: MP3, WAV, M4A, MP4, AVI, MOV, WebM, MKV
• YouTube Integration: Direct URL processing for YouTube videos
• AI-Powered Formatting: Clean, readable transcript formatting
• Chunked Processing: Handles large files by splitting into manageable chunks

🤖 AI Providers

• OpenAI: Whisper for transcription, GPT-4 for analysis (recommended for structured responses)
• Google Gemini: Alternative AI provider for basic formatting

Quick Start

Backend Setup

1. Install Dependencies

pip install fastapi uvicorn python-multipart openai google-generativeai yt-dlp ffmpeg-python python-dotenv

2. Environment Configuration Create a .env file:

OPENAI_API_KEY=your_openai_api_key_here
GEMINI_API_KEY=your_gemini_api_key_here  # Optional

3. Run the FastAPI Server

python main.py
# Server will start on http://localhost:8000

Frontend Setup

1. Navigate to Frontend Directory

cd frontend

2. Install Dependencies

npm install

3. Environment Configuration Create a .env.local file:

NEXT_PUBLIC_API_URL=http://localhost:8000

4. Run the Development Server

npm run dev
# Frontend will start on http://localhost:3000

API Endpoints

📊 Interview Analysis

POST /analyze-interview

Comprehensive interview analysis with skill assessment and insights.

Parameters:

• file: Audio/video file (multipart/form-data)
• skills_to_assess: Comma-separated skills list (form field)
• job_role: Job role for context (form field, default: "Software Developer")
• company_name: Company name for context (form field, default: "Company")
• ai_provider: AI provider (form field, default: "openai")

Example Request:

curl -X POST "http://localhost:8000/analyze-interview" \
  -F "file=@interview.mp3" \
  -F "skills_to_assess=Python,React,Problem Solving,Communication" \
  -F "job_role=Senior Developer" \
  -F "company_name=Tech Corp"

Response Structure:

{
  "filename": "interview.mp3",
  "raw_transcript": "...",
  "formatted_transcript": "...",
  "ai_provider": "openai",
  "file_chunks": 2,
  "skill_assessments": [
    {
      "skill": "Python",
      "level": "Advanced",
      "confidence_score": 85,
      "evidence": "Demonstrated strong knowledge...",
      "recommendations": "Consider exploring..."
    }
  ],
  "questions_and_answers": [
    {
      "question": "Tell me about your Python experience",
      "answer": "I have been working with Python...",
      "grade": "Good",
      "score": 78,
      "feedback": "Good technical explanation...",
      "key_points_covered": ["Experience", "Projects"],
      "areas_for_improvement": ["More specific examples"]
    }
  ],
  "interview_insights": {
    "overall_performance_score": 75,
    "communication_clarity": 80,
    "technical_depth": 70,
    "problem_solving_ability": 85,
    "confidence_level": 75,
    "strengths": ["Clear communication", "Strong technical knowledge"],
    "weaknesses": ["Could provide more examples"],
    "hiring_recommendation": "Recommend for hire with considerations...",
    "next_steps": ["Technical deep-dive interview", "Reference checks"]
  },
  "analysis_summary": "Executive summary of the analysis..."
}

POST /analyze-interview-url

Same as above but accepts a video URL instead of file upload.

Parameters:

• video_url: Video URL (YouTube, etc.)
• skills_to_assess: Comma-separated skills list
• job_role: Job role for context
• company_name: Company name for context
• ai_provider: AI provider

📝 General Transcription

POST /extract-transcript

Extract and format transcript from video URL.

POST /upload-audio

Upload audio file for transcription and formatting.

Skill Assessment Levels

• Expert: Demonstrates mastery and deep understanding
• Advanced: Strong knowledge with practical application
• Intermediate: Good understanding with some gaps
• Beginner: Basic knowledge, needs development
• Not Demonstrated: Skill not shown in the interview

Q&A Grading Scale

• Excellent (90-100): Outstanding answer with depth and clarity
• Good (70-89): Solid answer covering key points
• Average (50-69): Adequate answer with room for improvement
• Below Average (30-49): Weak answer missing key elements
• Poor (0-29): Inadequate or incorrect answer

Performance Metrics

The system evaluates candidates across multiple dimensions:

• Overall Performance: Comprehensive score (0-100)
• Communication Clarity: How well the candidate communicates
• Technical Depth: Level of technical knowledge demonstrated
• Problem Solving: Ability to think through problems
• Confidence Level: Candidate's confidence and self-assurance

Best Practices

For Accurate Analysis

1. Clear Audio Quality: Ensure good audio quality for better transcription
2. Specific Skills: List specific, relevant skills for assessment
3. Proper Context: Provide accurate job role and company information
4. Interview Structure: Well-structured interviews yield better analysis

Skill Input Examples

• Technical: "Python, JavaScript, React, SQL, AWS, Docker"
• Soft Skills: "Communication, Leadership, Problem Solving, Teamwork"
• Domain-Specific: "Machine Learning, Data Analysis, System Design"

Limitations

• Currently optimized for English-language interviews
• Works best with structured interview formats
• Requires OpenAI API for comprehensive analysis features
• Analysis quality depends on audio quality and interview content

Error Handling

The API includes comprehensive error handling:

• File format validation
• Transcript quality validation
• Skill count limits (max 20 skills)
• AI provider compatibility checks
• Structured response validation

Security Considerations

• Files are processed temporarily and cleaned up automatically
• No permanent storage of audio/video files
• API keys should be secured and rotated regularly
• Consider rate limiting for production deployments

Development

Adding New Analysis Features

1. Define new Pydantic models in main.py
2. Create analysis functions using OpenAI structured responses
3. Update the response models to include new data
4. Add frontend components to display new insights

Custom AI Prompts

The system uses carefully crafted prompts for each analysis type. You can customize these prompts in the respective functions to better suit your interview style or requirements.

Contributing

Contributions are welcome! Please ensure:

• Add appropriate error handling
• Include type hints and documentation
• Test with various file formats and interview styles
• Follow the existing code structure and patterns

License

MIT License - see LICENSE file for details.