Youtube Study Companion

An open-source YouTube companion bot that lets you query large YouTube playlists and videos in natural language.

Comment 2

📜 Overview

YouTube study companion, the idea is basically to give companionship, a guide to learners who study through lengthy YouTube videos and playlists so they can study their topics easily and efficiently instead of browsing through all those lengthy playlists just to find their topics of interest.

❓ Problem Statement

Professionals & Students encounter challenges when searching through lengthy playlists & YouTube videos for specific information or topics of interest.

🧩 Solution

Identifying and playing specific sections related to user queries & their topics of interest. Streamlined & simplify the process of querying & searching specific topics through our agentic flow.

⚙ How It Works

Users just need to provide the link to a lengthy YouTube video or playlist and let the bot train, after that users can ask questions to the bot and it will shortly return those time stamps where the topic has been discussed. Users can simply watch the video from those timestamps just by clicking on it.

💡 What I Learned

  • Efficient Use of similarity search.
  • Use of Generative AI in a more revolutionary manner than that of a simple RAG.
  • Usefulness of TiDB Cloud.

🔨🔧 Tech Stack

  • TiDB (vector similarity search)
  • iDB - MySQL
  • Jina AI - Embedding API
  • NLP through NLTK (for stopwords removal)
  • Web-app (React with NextJS & Django)

🚀 Future Plans

(described in the video)

  • Note Taking
  • Summarization
  • A study companion with multiple data sources.
  • Porting to a web extension.

📎Source Code

📎API Documentation

📎Presentation

Built With

Share this project:

Updates