📄 Scientific Paper Q&A Assistant (RAG)

A production-ready, cloud-deployed Retrieval-Augmented Generation (RAG) application designed to parse academic papers and scientific PDF publications, enabling users to extract accurate, context-aware answers with transparent source citations.

🚀 Live Demo: https://research-paper-q-a-assistant.streamlit.app/

📂 GitHub Repository: https://github.com/wayniez/Research-paper-Q-A-Assistant

---

🌟 Features

📑 PDF Document Processing

Upload and parse complex scientific papers with ease.

🧩 Semantic Chunking

Advanced text-splitting pipeline with sliding-window overlap to preserve context across document segments.

🔍 Vector Search Engine

Dense semantic retrieval powered by open-source embeddings and local vector indexing.

🤖 Context-Aware Question Answering

Fast LLM inference that generates answers exclusively from the uploaded document context.

✅ Source Attribution & Factuality

Each response is linked back to relevant document excerpts, improving transparency and reducing hallucinations.

---

🛠️ Tech Stack

Category	Technology
Language	Python 3.11
Frontend	Streamlit
Vector Database	ChromaDB
Embedding Model	Hugging Face all-MiniLM-L6-v2
LLM	LLaMA 3 (via Groq API)
Configuration	TOML
Version Control	Git

---

📂 Project Structure

├── app.py
├── rag_engine.py
├── vectorstore.py
├── requirements.txt
└── .gitignore

File Overview

File	Description
app.py	Streamlit UI and application entry point
rag_engine.py	Core RAG pipeline, chunking, retrieval, and LLM orchestration
vectorstore.py	Embedding generation and ChromaDB integration
requirements.txt	Project dependencies
.gitignore	Excludes local databases, virtual environments, and temporary files

---

⚡ Local Installation

1️⃣ Clone the Repository

git clone https://github.com/wayniez/Research-paper-Q-A-Assistant.git
cd Research-paper-Q-A-Assistant

2️⃣ Create a Virtual Environment

Linux / macOS

python -m venv venv
source venv/bin/activate

Windows

python -m venv venv
venv\Scripts\activate

3️⃣ Install Dependencies

pip install -r requirements.txt

Note: httpx==0.27.2 is intentionally pinned to ensure compatibility with the Groq SDK.

4️⃣ Configure Secrets

Create the following file:

.streamlit/secrets.toml

Add your Groq API key:

GROQ_API_KEY = "your_actual_groq_api_key_here"

5️⃣ Run the Application

streamlit run app.py

---

☁️ Deployment on Streamlit Cloud

This application is optimized for deployment on Streamlit Community Cloud.

Deployment Steps

1. Push the project to GitHub.
2. Connect the repository to Streamlit Cloud.
3. Set the main file path to:

app.py

4. Open Advanced Settings → Secrets.
5. Add your API key:

GROQ_API_KEY = "your_actual_groq_api_key_here"

6. Deploy the application.

---

🚀 How It Works

PDF Upload
     │
     ▼
Document Parsing
     │
     ▼
Semantic Chunking
     │
     ▼
Embeddings Generation
     │
     ▼
ChromaDB Vector Store
     │
     ▼
Similarity Search
     │
     ▼
Retrieved Context
     │
     ▼
LLaMA 3 (Groq)
     │
     ▼
Answer + Source Citations

---

📸 Application Workflow

• Upload a scientific paper (PDF).
• The document is parsed and split into semantic chunks.
• Chunks are embedded and stored in ChromaDB.
• User submits a question.
• Relevant context is retrieved from the vector database.
• LLaMA 3 generates an answer based solely on retrieved content.
• Supporting source passages are displayed alongside the answer.

---

🔒 Security

• API keys are stored securely using Streamlit Secrets.
• Local vector databases are excluded from version control.
• No document data is permanently stored in the repository.

---

📜 License

This project is licensed under the MIT License.

Feel free to fork, modify, and use it for personal or commercial projects.