🧠 AI Interview Prep Coach with Mem0 Memory

This project demonstrates how to build an AI agent with both short-term and long-term memory using Mem0, LangChain, and Groq. The agent acts as an interview preparation coach, remembering your answers and preferences across sessions.

---

Features

• Conversational UI: Chat with an AI interview coach in your browser (Streamlit app)
• Personalized Memory: Remembers your answers, feedback, and preferences across sessions
• Memory Engine: Uses Mem0 for memory extraction, storage, and retrieval (vector DB: Qdrant)
• LLM Integration: Uses Groq (Llama-3) for interview questions and Gemini for embeddings
• Input & Output validation: Checking the input before passing it to the LLM and processing the output before giving harmful response to the user

---

How It Works

1. User interacts with the AI via a chat interface (Streamlit)
2. Messages are sent to the memory engine (Mem0), which extracts and stores relevant memories
3. Memories are retrieved and provided as context to the LLM (Groq/Llama-3) for personalized responses
4. All conversations are stored in a vector database (Qdrant) for long-term recall

Memory Types

• Short-term memory: Session-based, remembers the current conversation
• Long-term memory: User-based, persists across sessions (using user_id)

---

Project Structure

memory_in_agents/
├── app.py                # Streamlit UI for chat
├── memory_agent.py       # Core logic: memory engine, LLM chain, chat function
├── MEMORY.md             # Theory and concepts of memory in AI agents
├── pyproject.toml        # Python dependencies
├── notebooks/
│   └── experiment.ipynb  # Example notebook for experiments
└── README.md             # This documentation

Core files involved are app.py and memory_agent.py for implementation purpose and experiement.ipynb for experiment

---

Setup & Installation

1. Clone the repo

   git clone https://github.com/Jerry-britto/AI-Interview-Coach.git
   cd AI-Interview-Coach

2. Create a virtual environment

   python3 -m venv .venv
   source .venv/bin/activate

3. Install dependencies

   pip install -r requirements.txt
   # or
   pip install -e .

4. Set up API keys

   • Copy .env.example to .env and add your keys for GROQ_API_KEY and GEMINI_API_KEY.

---

Running the App

Streamlit Web UI

streamlit run app.py

Open http://localhost:8501 in your browser.

CLI Demo

python main.py

---

Main Files Explained

• app.py: Streamlit web interface for chatting with the AI coach. Handles user input, session state, and displays stored memories.
• memory_agent.py: Implements the memory engine using Mem0, integrates with Groq LLM and Gemini embedder, and provides the chat() function.
• notebooks/experiment.ipynb: Jupyter notebook for experimenting with memory and retrieval.

---

How Mem0 Memory Engine Works

1. Receives user messages and extracts memories using an LLM
2. Resolves conflicts (e.g., updates preferences if changed)
3. Stores memories in a vector DB (Qdrant)
4. Retrieves relevant memories for each new query
5. Combines graph and vector search results for best context

See MEMORY.md for a deep dive into memory concepts and parameters.