personal Featured

Text Summarizer App

AI-powered text summarizer with file upload, analytics dashboard, and advanced LLM controls - built with Streamlit, OpenAI GPT-4, and LangChain

View Code
Python Streamlit OpenAI GPT-4 LangChain Pandas

Project Overview

AI-powered text summarizer with file upload, analytics dashboard, and advanced LLM controls - built with Streamlit, OpenAI GPT-4, and LangChain

README.md
README.md

Project Overview

An AI-powered text summarization application that leverages OpenAI’s GPT-4 through LangChain to provide intelligent document summaries. Features a Streamlit-based UI with file upload capabilities, analytics dashboard, and configurable LLM parameters for fine-tuned outputs.

Key Features

File Upload & Processing

  • Support for multiple document formats (TXT, PDF, DOCX)
  • Large file handling with chunked processing
  • Text extraction and preprocessing pipeline

AI-Powered Summarization

  • GPT-4 integration via LangChain
  • Configurable summary length and style
  • Multiple summarization modes (extractive, abstractive, bullet points)
  • Token usage tracking and cost estimation

Analytics Dashboard

  • Summary statistics (compression ratio, reading time saved)
  • Token usage history and trends
  • Export functionality for summaries and reports

Advanced Controls

  • Temperature adjustment for creativity control
  • Max token limits
  • Custom prompt templates
  • Model selection (GPT-3.5 vs GPT-4)

Technical Implementation

Technologies Used

  • Python 3.11+: Core development language
  • Streamlit: Interactive web interface
  • OpenAI GPT-4: Language model for summarization
  • LangChain: LLM orchestration and chain management
  • Pandas: Data processing and analytics
  • python-docx/PyPDF2: Document parsing

Architecture

  1. Document Ingestion: File upload and text extraction
  2. Preprocessing: Text cleaning and chunking
  3. LLM Chain: LangChain pipeline for summarization
  4. Analytics Engine: Metrics calculation and history tracking
  5. Streamlit UI: Interactive dashboard and controls

Challenges & Solutions

  1. Large File Handling: Implemented chunked processing to handle documents exceeding token limits
  2. Cost Management: Built token tracking to monitor API usage and estimate costs
  3. UI Responsiveness: Added progress indicators and async processing for better UX

Learnings

  • LangChain’s chain composition for complex LLM workflows
  • Streamlit’s session state management for multi-page apps
  • Balancing LLM quality vs cost (GPT-4 vs GPT-3.5 tradeoffs)
  • Document parsing edge cases and encoding issues
Back to all projects