Arabic PDF Conversions Using Pipeline & RAG

We are developing a data ingestion pipeline and RAG (Retrieval-Augmented Generation) system to process, extract, structure, and index content from Arabic PDF files. The goal is to enable users to search and ask questions solely from provided book or file content.

Key Requirements:
- Process one chapter from an Arabic PDF as a test case.
- Extract and structure both structured and non-structured data using Python.
- Store, index, and enable semantic and vector search capabilities.
- Enable search by text, voice, video, and image inputs.
- Answers must strictly come from indexed data with source references; unsupported answers should indicate "not available." No hallucinations or guesses allowed.

Deliverables:
- Full pipeline implementation for structured and non-structured ingestion of Arabic PDF content.
- Search-ready dataset with proper indexing techniques.
- Semantic search capability compatible with text, voice, video, and image queries with concise, referenced answers or no-answer responses when applicable.

Timeline: 3 days.
Preferred Tech Stack: Python. Expert knowledge of data pipelines and RAG systems is required.