🤖 MATLAB Knowledge Assistant

AI Knowledge Assistant

This project is an AI-powered knowledge assistant designed to help users troubleshoot MATLAB-related issues using retrieval-augmented generation (RAG), multi-agent workflows, and large language models. The system combines LangGraph, FAISS, Gemini, and document retrieval techniques to provide contextual and source-grounded responses through an interactive web interface.

🛠️ Tech Stack

• Frontend: Streamlit
• Backend: LangChain,Langgraph, Python, Flask
• LLMs: Hugging Face Transformers, Gemini
• Vector Store: FAISS
• Database: MongoDB

---

🚀 Features

• Chat-based troubleshooting for MATLAB errors and syntax
• Chat-based multi-turn troubleshooting with memory.
• Image based trouleshooting for MATLAB errors.
• Citation of sources or documentation sections in responses.
• Analytics dashboard for admin users.
• Contextual query understanding via document retrieval.
• LLM-powered answers based on MATLAB full documentation.
• User-friendly UI interface.
• Multi Chat Sessions maintainenece to access previous chats
• Authentication for user and admin accounts.
• Modular, extensible backend pipeline.

---

Project Structure

├── backend
│   ├── agents
│   │   ├── answerQnaAgent.py
│   │   ├── answerRagAgent.py
│   │   ├── autocompleteAgent.py
│   │   ├── decisionAgents.py
│   │   ├── imageQueryAgent.py
│   │   ├── intialAnsweringAgent.py
│   │   ├── qnaDbAgents.py
│   │   ├── queryAnnotatorAgent.py
│   │   └── scrapingAgent.py
│   ├── database.py
│   ├── faiss_vector_store
│   │   ├── index.faiss
│   │   └── index.pkl
│   ├── main.py
│   ├── __pycache__
│   │   └── main.cpython-310.pyc
│   ├── qnaDB
│   │   ├── index.faiss
│   │   └── index.pkl
│   └── requirements.txt
├── Backend.jpg
├── frontend
│   ├── app.py
│   └── libs
├── Frontend_UI.jpg
├── README.md
├── results
└── visited.txt

🧩 Frontend User Flow

---

🔧 Backend Flow

---

⚙️ Setup Instructions

.env file example (in backend directory) -

GEMINI_API_KEY=your_gemini_api
MONGODB_URI=your_database_uri
HUGGINGFACEHUB_API_TOKEN="your_huggingface_api_token"

1. Clone the repository:

git clone https://github.com/your-username/matlab-chatbot.git
cd matlab-chatbot

2. Install dependencies:

pip install -r requirements.txt

3. Run the app:

python backend/database.py

streamlit run frontend/app.py

📄 Example Queries

Examples queries

Future Improvements

• Support for custom document uploads
• Advanced reranking techniques
• Expanded multi-domain knowledge bases
• Improved citation generation
• Containerized deployment using Docker