🤖 AskMyDocs

An AI-powered chatbot that answers questions from your PDF documents using RAG (Retrieval Augmented Generation) architecture, powered by LangChain + Groq LLaMA3 + Streamlit.

Upload any PDF → Ask questions → Get answers strictly from your documents.

---

🏗️ Architecture (RAG Pipeline)

Upload PDF from UI
      ↓
PyPDFLoader (LangChain) extracts text
      ↓
PDF content → Context string
      ↓
PromptTemplate (context + chat history + question)
      ↓
Groq LLaMA3-70b (LLM)
      ↓
StrOutputParser → Answer displayed in Streamlit UI

---

✨ Features

• Upload multiple PDFs directly from the UI sidebar
• Context-bound answers — LLM only uses PDF content
• Chat history maintained across conversation
• Cached LLM instance — fast responses
• Clean Streamlit chat UI
• Clear chat history button
• Graceful error handling

---

🛠️ Tech Stack

Layer	Technology
LLM	Groq API (LLaMA3-70b-versatile)
RAG Framework	LangChain
Document Loading	PyPDFLoader + tempfile
Prompt	LangChain PromptTemplate
Chain	LCEL (LangChain Expression Language)
Frontend	Streamlit

---

🚀 Setup & Run

1. Clone the repo

git clone https://github.com/Prajwal4581/AskMyDocs.git
cd AskMyDocs

2. Create virtual environment

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

3. Install dependencies

pip install -r requirements.txt

4. Set up API key

cp .env.example .env
# Add your Groq API key in .env
GROQ_API_KEY=gsk_xxxxxxxxxxxxxxx

Get free Groq key → https://console.groq.com

5. Run the app

streamlit run app.py

6. Open browser → http://localhost:8501

• Upload PDFs from sidebar
• Ask questions in chat
• Get answers from your documents 🎯

---

📁 Project Structure

AskMyDocs/
├── app.py              # Main Streamlit app + RAG chain
├── requirements.txt
├── .env.example
├── .gitignore
└── README.md

---

📸 Sample Usage

Question	Source
"What are the tables in Residentia?"	Residentia PDF → lists all 12 tables
"What is Spring Boot?"	Java notes PDF → explains concept
"Summarize this document"	Any PDF → gives summary

---

🔮 Production Improvements

• Replace PDF concatenation with FAISS/ChromaDB vector store
• Add RecursiveCharacterTextSplitter for large documents
• Add semantic embeddings for better retrieval accuracy
• Support DOCX, TXT file formats
• Deploy on Streamlit Cloud with shareable URL
• Add source citation (which page answered the question)