AI Research Assistant

A portfolio-ready AI Research Assistant built with LangChain, Google Gemini API, and FAISS vector database.
Upload multiple PDFs and ask questions — the AI retrieves answers from your documents, with a collapsible “View Sources” feature under each response.

---

Features

• Multi-PDF upload
• Chat-style interface (like ChatGPT)
• Collapsible and styled “View Sources” with PDF name and page
• RetrievalQA powered by Google Generative AI embeddings
• Multi-turn conversation with history
• FAISS vectorstore persistence for faster repeated queries
• Clean, professional UI with Gradio

---

Installation

1. Clone the repository

git clone https://github.com/your-username/AI-Research-Assistant.git
cd AI-Research-Assistant

2. Create and activate a virtual environment

python -m venv .venv
source .venv/bin/activate   # Linux / macOS
.venv\Scripts\activate      # Windows

3. Install dependencies

pip install -r requirements.txt

4. Set up Google Application Default Credentials (ADC)

Follow Google Cloud ADC setup. Make sure you have a .env file in the project root with your API key if required:

GEMINI_API_KEY=your_google_gemini_api_key_here

---

Running the App

python main.py

• Upload one or more PDFs.
• Ask questions in the chat.
• Click View Sources under answers to see which PDF/page the answer came from.
• Clear chat with the Clear Chat button.

---

Project Structure

Albot/
├── main.py                # Gradio app + Chat interface
├── utils.py               # Document loading, embeddings, vectorstore
├── requirements.txt       # Dependencies
├── .env                   # API keys (ignored in git)
├── vectorstore/           # FAISS vector database
└── .venv/                 # Python virtual environment

---

Notes

• Do not commit .env or .venv. They are included in .gitignore.
• The vectorstore folder contains embeddings and is also ignored.
• FAISS deserialization is safe for your own data but should not be loaded from untrusted sources.

---

License

MIT License.