📄 AskYourPDF — AI Document Question-Answering (Capstone Project)

A simple and clean AI-powered PDF Question Answering Agent built as part of the 5-Day AI Agents Intensive Course with Google & Kaggle (2025) where users upload a PDF and ask questions. This repository is a minimal, demonstrable project that satisfies the Kaggle/Google Capstone submission requirements.

📸Screenshots

1. Home Page:

---

2. Results Page:

---

3. Architecture:

---

Users can:

• Upload a PDF

• Ask questions about its contents

• Receive contextual answers with source references

• (Optional) Run in mock embedding mode to avoid API costs / quota usage

---

This project is intentionally minimal, easy to run, and perfect for a capstone submission.

---

✨ Features

🔹Upload any PDF (up to ~10MB)

🔹Background processing of chunks

🔹Mock Embeddings Mode → works even without OpenAI API key

🔹Ask natural language questions about the document

🔹Source pages + similarity scores

🔹Beautiful and responsive React UI

🔹Node.js backend with in-memory vector search

---

🖥️ Tech Stack

Frontend

• React (Vite)

• Custom UI (no CSS frameworks needed)

• Fetch API for server communication

Backend

• Node.js + Express

• PDF parsing (pdf-parse)

• In-memory vector store

• OpenAI-compatible structure (mock or real)

• Multer for file upload

---

📂 Folder Structure

ask-your-pdf
├── client/         # React frontend
├── server/         # Node.js backend
│   ├── embeddings.js
│   ├── index.js
│   └── uploads/
├── README.md       # This file
└── ...

---

🔧 Installation & Setup

Clone the repository:

git clone https://github.com/yourusername/ask-your-pdf.git
cd ask-your-pdf

---

🛠️ Backend Setup (Server)

cd server
npm install

---

Environment variables

Create a .env file:

OPENAI_API_KEY=

Start the backend:

npm run dev

---

🎨 Frontend Setup (Client)

cd ../client
npm install
npm run dev

---

📚 How It Works

1. Upload PDF

   The server extracts text, splits it into chunks, and (in real mode) generates embeddings.

2. Ask a Question

   The question is embedded and compared with stored chunk vectors.

3. Relevant chunks are fed into the model

   In mock mode → uses local pseudo-embeddings

   In real mode → uses OpenAI embeddings + Chat Completions

4. Final answer is displayed with source references.

---

🧪 Mock Embeddings Explained

This project supports a full offline mode where:

• No API calls are made

• No cost is incurred

• Embeddings are simulated using deterministic vectors

---

🌟 Credits

Built by Kamran (mohdkamrankhan) Capstone for Google x Kaggle — 5-Day AI Agents Intensive (2025)

---

📄 License

This project is open-source and available under the [MIT License]

---

📧 Contact / Feedback

If you find any bugs or want to suggest improvements, feel free to open an issue or create a pull request. You can reach me at: [mohdkamrankhan.dev@gmail.com]

---

💡 If you like my work, please ⭐ my repos. Your support inspires me to build more projects! 🚀

---

👨🏻‍💻 From mohdkamran-khan