AI Interview Curator

A Streamlit-based AI interview preparation tool that analyzes a Job Description and a resume to generate:

• ✔️ Candidate fit summary
• ✔️ Good-fit & weak-fit breakdown
• ✔️ Tailored technical questions (with difficulty + follow-ups)
• ✔️ Tailled behavioral questions (with follow-ups)
• ✔️ Optional expected answer outlines
• ✔️ Clean Format-B structured output
• ✔️ Runs locally using Ollama (no cloud, fully private)

Built for students, job-seekers, and engineers who want realistic, role-specific interview prep generated entirely on their machine.

---

How It Works

1. Upload or paste:

   • A Job Description
   • Your resume (PDF, DOCX, or TXT)

2. Select:

   • Number of technical questions
   • Number of behavioral questions
   • Difficulty level
   • Whether to include expected answers

3. Click Generate
   The app uses Ollama (local LLMs) to produce a crisply formatted interview-prep document.

---

🛠️ Tech Stack

• Python
• Streamlit (UI)
• Ollama (local model inference)
• LLMs (e.g., Llama 3, Qwen, Mistral, etc.)
• PyPDF2 / python-docx for file parsing

---

📁 Project Structure

ai_interview/
│
├── app.py                    # Main Streamlit app (thin UI layer)
│
├── config.py                 # Constants & defaults
│
├── loaders/
│   ├── __init__.py
│   └── file_loader.py        # PDF / DOCX / TXT extraction
│
├── llm/
│   ├── __init__.py
│   ├── prompt_builder.py     # Prompt construction
│   └── ollama_client.py      # Ollama HTTP / CLI calls
│
├── tts/
│   ├── __init__.py
│   └── speaker.py            # TTS logic
│
├── interview/
│   ├── __init__.py
│   └── question_parser.py    # Extract first question
│
└── requirements.txt

---

⚙️ Installation

1. Install Python packages

pip install streamlit PyPDF2 python-docx requests

2. Install Ollama

Download from:
https://ollama.com/download

3. Pull a model (example)

ollama pull llama3

4. Run the app

streamlit run Ai_Interview_Prep.py

---

Features

Candidate Fit Summary

• Highlights strengths across skills, tools, and experience
• Identifies gaps or mismatches

Technical Questions

Each question includes:

• Difficulty
• Follow-up question
• (Optional) expected answer outline

Questions are fully tailored to the JD & resume.

Behavioral Questions

• STAR-style follow-ups
• Role-specific behavioral prompts

100% Local

Your resume and JD never leave your machine.
Ollama handles everything locally.

---

Example Use Cases

• Preparing for software engineering, data, ML, DevOps, or PM interviews
• Getting a fast “mock interview prep sheet” from any job posting
• Practicing with role-specific, realistic questions
• Reviewing job fit and skill gaps

---

Customize It

Want to add:

• PDF export
• Difficulty color badges
• Custom question categories
• Multiple model support
• STAR-format answer breakdowns

Just ask! The project is easy to extend.

---

Contributing

PRs welcome!
You can contribute by:

• Adding new question styles
• Improving prompts
• Supporting additional file types
• Enhancing UI/UX

---

x