This project contains our solution for the Adobe India Hackathon 2025, a comprehensive suite of tools designed to transform how we interact with PDFs.
-
Challenge 1A: PDF Heading Extractor We built a fast, accurate, and multilingual system to extract the structural outlines of PDF documents. Our hybrid approach uses both fast heuristic analysis and a lightweight AI model to identify and classify headings (Title, H1, H2, H3), delivering reliable results in seconds.
-
Challenge 1B: Persona-Driven Document Intelligence We developed an intelligent system that analyzes a collection of PDFs to find the most relevant information for a user's specific job. The solution uses a sentence transformer model to prioritize sections based on a given persona and task, generating a structured output for easy consumption.
You can find the full documentation and code in the respective challenge folders below.
Made by @ivar03 and @exploring-solver.