Inspiration

Cooking tutorials on YouTube are abundant, but it’s often hard to keep track of the exact steps, ingredients, or tips shown in a video. Many times, you find yourself rewinding just to remember the right moment. We wanted to solve this by creating a tool that makes cooking videos actionable — allowing users to annotate, tag, and quickly revisit important parts.


What it does

Cooktube is a web app that helps users organize and enhance YouTube cooking videos:

  • Save and manage your favorite cooking videos.
  • Add custom tags and edit video titles for quick categorization.
  • Take timestamped notes so you can jump back to specific steps (e.g., “add garlic at 2:15”).
  • Search videos by tags.
  • Use AI to automatically extract ingredients from subtitles and convert them into searchable tags.

How we built it

  • Frontend: React for a responsive and user-friendly interface.
  • Backend: FastAPI with Python for handling video and note management.
  • Database: SQLite for lightweight, easy-to-deploy storage.
  • AI: Hugging Face models and OpenAI’s GPT-OSS-120B for extracting ingredients from transcripts.
  • Deployment: Frontend on Vercel, Backend on Render, making it easy to scale and share.

Challenges we ran into

  • Handling YouTube transcript errors (like HTTP 426 responses).
  • AI token limits when analyzing long transcripts.
  • Syncing annotations smoothly with video playback.
  • Balancing a fast prototype with reliable user authentication (Google OAuth).

Accomplishments that we're proud of

  • Built a functional end-to-end system in a short time frame.
  • Integrated AI to turn raw transcripts into useful ingredient tags.
  • Created a clean, intuitive UI that makes video organization simple and practical.
  • Successfully deployed both frontend and backend to production.

What we learned

  • How to combine AI + video content for a better user experience.
  • The importance of designing around API limitations (YouTube transcripts & AI token constraints).
  • Best practices for deploying lightweight full-stack apps using Vercel and Render.
  • How small UX touches (like timestamped notes) can make a big impact on usability.

What's next for Cooktube

  • Add support for collaborative recipe sharing (friends can share their tagged/annotated videos).
  • Improve AI ingredient extraction to support multi-language subtitles.
  • Build mobile-first features for on-the-go cooking.
  • Enhance search with advanced filters (by ingredient, cuisine, difficulty).
  • Integrate grocery list generation directly from video ingredients.

Built With

Share this project:

Updates