Home / About
About

About PatternPulse AI

Learn about the technology behind intelligent document classification and secure PDF redaction.

What is PatternPulse AI?

PatternPulse AI is an intelligent document processing platform designed for the Architecture, Engineering, and Construction (AEC) industry. It uses machine learning to automatically classify construction drawings into six engineering disciplines: Architectural, Structural, Mechanical, Plumbing, Electrical, and Fire Protection.

Beyond classification, PatternPulse AI provides secure PDF redaction capabilities using Natural Language Processing (NLP) to detect and permanently remove sensitive information such as names, email addresses, phone numbers, physical addresses, and project identifiers from construction documents.

Project Architecture

PatternPulse AI follows a modern client-server architecture with a React frontend and FastAPI backend, connected through a RESTful API.

React Frontend User Interface
→
FastAPI Backend REST API
→
ML Pipeline scikit-learn + spaCy
→
PyMuPDF PDF Processing

AI Pipeline

Documents flow through a multi-stage pipeline that extracts text, classifies content, identifies sensitive data, and generates redacted outputs.

1Upload PDF Document
↓
2Extract Text via PyMuPDF
↓
3Classify with ML Model
↓
4Detect PII with spaCy NER
↓
5Generate Redacted PDF
↓
6Download Results & Report

Tech Stack

PatternPulse AI is built using a modern, production-ready technology stack optimized for performance, scalability, and developer experience.

  • React — Interactive frontend with component-based architecture
  • FastAPI — High-performance Python backend with async support
  • Python — Core language for ML pipeline and data processing
  • scikit-learn — Machine learning model for document classification
  • spaCy — NLP-based named entity recognition for PII detection
  • PyMuPDF — Fast PDF text extraction and redaction rendering
  • Tailwind CSS — Utility-first styling for rapid UI development

FAQ

Currently, PatternPulse AI supports PDF files only. Support for DOCX, DWG, and DXF formats is planned for future releases.
The classification model achieves an average accuracy of 94.2% across all six engineering disciplines. The model is continuously improving as more training data is collected.
PatternPulse AI can detect and redact person names, email addresses, phone numbers, physical addresses, and project identifiers using a combination of spaCy NER and regex pattern matching.
Yes. All uploaded files are encrypted during transit and at rest. Files are automatically deleted within 24 hours of processing. PatternPulse AI is fully GDPR compliant.
Yes. PatternPulse AI exposes a RESTful API built on FastAPI. You can integrate classification and redaction into your own workflow. See the documentation page for details.