📄 Enterprise Document Classifier Service

An enterprise-ready document classification service built with FastAPI that automatically classifies uploaded documents and recommends the optimal processing strategy for Enterprise AI and RAG applications.

Features

• Upload documents through REST API
• Automatic document classification
• Rule-based classifier
• Optional LLM-powered classifier
• OCR detection
• Chunking strategy recommendation
• Metadata recommendation
• Processing pipeline recommendation
• Table extraction recommendation
• Enterprise AI ready

Supported Document Types

• HR Policies
• Legal Contracts
• Technical Documentation
• Research Papers
• Source Code
• FAQ / Knowledge Base
• Invoices
• General Documents

Recommended Chunking Strategies

Document Type Recommended Strategy

---

Policy Documents Section-based Chunking Legal Contracts Section + Semantic Chunking Technical Documentation Heading-based Chunking Source Code Function/Class Chunking Research Papers Semantic Chunking FAQ Documents Question-Answer Chunking CSV / Excel Table Chunking General Documents Fixed-size Chunking

Processing Pipeline

Enterprise Documents
        │
        ▼
Text Extraction
        │
        ▼
OCR Detection
        │
        ▼
Document Classification
        │
        ▼
Metadata Recommendation
        │
        ▼
Chunking Strategy Recommendation
        │
        ▼
Table Extraction Decision
        │
        ▼
Chunk Generation
        │
        ▼
Embedding Generation
        │
        ▼
Vector Database

Tech Stack

• FastAPI
• Python
• OpenAI API (optional)
• Pydantic

Installation

git clone https://github.com/<your-username>/document-classifier.git
cd document-classifier
python -m venv venv
source venv/bin/activate
pip install -r requirements.txt

Run

uvicorn app.main:app --reload

Open:

http://127.0.0.1:8000/docs

Sample Response

{
  "document_type": "Policy / HR Document",
  "recommended_chunking": "Section-based Chunking",
  "requires_ocr": false,
  "extract_tables": false
}

Roadmap

• Layout Detection
• Table Detection
• Local LLM Support
• Batch Processing
• Vector Database Integration
• Evaluation Dashboard

License

MIT

Author

Jay Ram Singh

Building Enterprise AI Platforms, RAG Systems, Document Intelligence, and LLMOps.