Back to Projects
RAG Chatbot
CompletedPythonStreamlitOpenRouter API+4 more

RAG Chatbot

A Streamlit-based chatbot that lets users upload PDFs and ask context-aware questions using retrieval-augmented generation.

Timeline

7 days

Role

Full Stack

Team

Solo

Status
Completed

Technology Stack

Python
Streamlit
OpenRouter API
scikit-learn
PyPDF2
NumPy
python-dotenv

Key Challenges

  • PDF Text Extraction Quality
  • Context Retrieval Accuracy
  • Rate Limiting
  • Model Response Handling
  • Session State Management

Key Learnings

  • RAG Fundamentals
  • Vector Similarity Search
  • Prompt Structuring
  • Streamlit State Management
  • API Error Handling
  • Production Deployment

RAG Chatbot: Chat with Your PDFs

Overview

RAG Chatbot - A lightweight application to upload PDF files and ask natural-language questions grounded in document content.

What Users Can Do

  • File Upload: Upload one or multiple PDF files and process them into searchable chunks.
  • Question Answering: Ask questions and receive answers generated from the most relevant document context.
  • Conversation Continuation: Continue conversations with reply-style follow-ups while keeping context in session history.

Why I built this

I built this platform to solve a fundamental issue I faced while studying as follows:

  • Reading long PDFs manually is slow and makes information retrieval inefficient.
  • It is difficult to quickly find precise answers across multiple documents during revision.

Tech Stack

Python Streamlit OpenRouter API scikit-learn PyPDF2 NumPy python-dotenv

After launch & Impact

  • Made document Q&A faster and easier for personal study and quick reference.
  • Deployed a working public app on Streamlit Cloud for easy access and sharing.
  • Improved understanding of end-to-end RAG flow from ingestion to retrieval to generation.
  • Built practical experience in model integration, rate limiting, and user-session handling.

Future Plans

  • Improve document chunking and retrieval quality for more accurate answers.
  • Add support for more file formats and richer citation-style responses.
  • Introduce configurable model and retrieval settings for advanced users.

anvesh.dev

© 2026 Anvesh Mishra. All rights reserved.