AI Content Transformation Platform - Backend

Backend service for the AI-powered content transformation platform developed for Smart India Hackathon 2026.

Overview

This backend accepts source content such as text, documents, presentations and images, applies user-defined transformation parameters, uses AI to generate the requested communication artefact, and returns the generated output in the selected format.

Architecture

Frontend ↓ FastAPI ↓ Input Processing (text / PDF / DOCX / PPTX / image) ↓ Unified Source Text ↓ Prompt Builder ↓ LLM (Groq) ↓ Generated Content ↓ Output Generator ↓ PDF / DOCX / PPTX / PNG

Features

• Text-based content transformation
• PDF, DOCX and PPTX document processing
• Image content understanding (extracts text and visual context from uploaded images)
• Configurable target audience
• Configurable communication objective
• Configurable tone
• Configurable language
• Configurable detail level
• Configurable content style
• Multiple output types, including social media formats (LinkedIn, Twitter/X, Instagram)
• PDF, DOCX, PPTX and PNG output
• AI-powered content generation using Groq
• Image understanding using Gemini vision
• AI-generated visual output using Hugging Face image generation

Tech Stack

• Python
• FastAPI
• Uvicorn
• Groq API (GPT-OSS 120B)
• Google Gemini API (vision)
• Hugging Face Inference API (image generation)
• PyPDF
• python-docx
• python-pptx
• ReportLab
• python-dotenv

Project Structure

backend/
├── main.py
├── config.py
├── instructions.py
├── input_processor.py
├── ai_engine.py
├── output_generator.py
├── gemini_engine.py
├── requirements.txt
├── .env
├── .gitignore
└── README.md