Chat with PDF - Technical Test

You are required to design a Chat with PDF system for interacting with PDF documents. The system should allow users to upload PDF files and interact with the content using AI-powered chat functionality.

Requirements:

Your system must include the features listed in Must have. You can also implement features in Nice to have for bonus points.

Must have

1. PDF Upload:
   • Users can upload PDF files.
   • Support scanned PDFs
   • Note: The PDF files may contain images, tables, and code.
2. AI Chat Functionality:
   • Allow users to ask questions.
   • Differentiate:
     • Document-based questions: Use content from document to generate answer.
     • General knowledge questions: Use an external AI API to provide responses unrelated to the document.
3. Demo Using Streamlit or Gradio (video or a direct link to the application):
   • Build a simple web-based interface to demonstrate the application.
   • Include PDF upload and chat functionalities in the demo.

Nice to have

• Multiple Files (PDFs, doc,...): Allow users to upload and interact with multiple files simultaneously.
• Custom UI: Create a simple custom UI (HTML/CSS/JS or frontend framework) to demo the application.

Technical requirements

• Language: Python
• Framework: Langchain, Llama-index, CrewAI, AutoGen, Haystack,…
• VectorDatabase: You can use any database you are familiar with such as: Milvus, Qdrant, Chroma, ….

How to Submit

• Duration: Complete the task within 3 days.
• Submission: Push your project to a public GitHub repository.
• Include:
  • A README.md file with:
    • Project overview.
    • Setup and installation instructions.
    • How to run the application.
    • How to test the application.
• Send the GitHub repository link via the submission form

---

Good luck!