🤖 AI-Powered Interview Coach

An interactive interview practice app built with Streamlit, Gemini API, and Docker. The app generates role-specific interview questions and provides real-time AI feedback on your answers.

✨ Features

• 🎯 Role-based interview question generation

• 📈 Adjustable difficulty level (Easy, Medium, Hard)

• 💬 Real-time AI feedback on answers

• 📦 Dockerized for easy deployment

• 🔄 Option to run with Gemini API (cloud) or Ollama (local LLM)

🛠️ Tech Stack

• Frontend/UI → Streamlit

• LLM → Google Gemini API (default) or Ollama (local option)

• Containerization → Docker + Docker Compose

• Language → Python 3.10+

📦 Installation

Clone the repo:

git clone https://github.com/izudada/interview_prep.git
cd interview-coach

Create a virtual environment:

python3 -m venv venv
source venv/bin/activate

Install dependencies:

pip install -r requirements.txt 

🔑 Environment Variables

Create a .env file in the root directory and add:

# For Gemini
GEMINI_API_KEY=your_gemini_api_key_here

# For Ollama (optional local model setup)
OLLAMA_HOST=http://localhost:11434

🚀 Running Locally

Option A: Using Gemini (default)

Run the docker container using:

make up

App will be live at http://localhost:8502

Option B: Using Ollama (local LLM)

Check into the containerization_branch:

git pull
git checkout containerization_branch

Then run the docker container using:

make up

🖥️ Usage

1. Select difficulty level and role in the sidebar.

2. Click Start Interview.

3. Answer questions in the text box.

4. Get personalized feedback in real-time.

5. Reset anytime to start fresh.

🏗️ Architecture

• Streamlit app (UI + logic)

• Gemini API or Ollama (LLM backend)

• Docker (containerization for deployment consistency)

📊 Performance & Scalability

• Gemini API → Google handles scaling; subject to rate limits.

• Ollama (local) → Performance depends on your machine’s CPU/RAM.

• Load testing → You can use Locust or ab to simulate 50+ concurrent users.

🔮 Roadmap

• Add voice input/output

• Save interview history

• Export feedback to PDF/CSV

• Multi-user support with database integration

📚 Resources

• Streamlit Docs

• Google Gemini API Docs

• Ollama Docs

• Docker Docs

• Locust Load Testing

• Dataquest - Prompting Large Language Models in Python