CareerCoachAI

An AI-powered interview coaching chatbot that remembers past conversations and gives smart, context-aware follow-up answers using Claude 3.5 and Cohere embeddings.

---

# CareerCoachAI

**CareerCoachAI** is an intelligent AI-powered chatbot that helps you prepare for technical job interviews. It remembers your past questions, retrieves relevant previous answers using semantic similarity, and gives you context-aware follow-up responses—just like a real interview coach.

---

## 🚀 Features

- 🧠 **Memory-based Interview Coaching**  
  Stores and recalls previous conversations for smarter interactions.

- 🤖 **Powered by Claude 3.5 Sonnet**  
  Uses the latest Anthropic model via Langchain for accurate and natural responses.

- 🔍 **Semantic Recall with Cohere Embeddings**  
  Embeds and compares user queries for smart similarity-based memory lookup.

- 💬 **Streamlit Frontend**  
  Clean, interactive interface for easy chat experience.

- 🔧 **Modular Python Architecture**  
  Clean separation of UI, logic, memory, and embeddings for easy extension.

---

## 📦 Installation

### 1. Clone the Repository

```bash
git clone https://github.com/your-username/CareerCoachAI.git
cd CareerCoachAI

2. Setup .env File

Create a .env file in the root directory with the following:

ANTHROPIC_API_KEY=your_anthropic_api_key
COHERE_API_KEY=your_cohere_api_key

3. Install Dependencies

pip install -r requirements.txt

4. Run the App

streamlit run streamlit_app.py

---

🗂️ Project Structure

CareerCoachAI/
├── streamlit_app.py         # Streamlit app interface
├── chatbot.py               # Chat logic and prompt invocation
├── chat_utils.py            # Load/save chat history
├── embedding_utils.py       # Cohere embedding + similarity matching
├── history.json             # Persistent chat history
├── embeddings.json          # Embedded Q&A for similarity recall
├── requirements.txt
├── README.md
└── .env                     # API keys (not tracked by Git)

---

🔮 Future Enhancements

• Use FAISS or ChromaDB for scalable vector search
• Add resume-based Q&A and context awareness
• Topic tagging for strengths/weakness identification
• Session-level tracking or multi-user support
• Dashboard for interview prep analytics

---

🧠 Built With

• Streamlit
• Langchain
• Anthropic Claude 3.5 Sonnet
• Cohere Embeddings
• Scikit-learn

---

🛡 License

This project is licensed under the MIT License.

---

👤 Author

\Padma Dey