DEV Community

Patel Rogers
Patel Rogers

Posted on

How I Built a Search Engine for MrBeast Videos using Python & Supabase

I realized something recently: MrBeast has uploaded over 700 videos, but there is no easy way to search through what he actually says.

If you want to find that one specific time Chandler won a challenge, or every time Jimmy mentioned "Feastables" in 2019, you have to manually scrub through hours of footage.

So, I spent the last weekend building a tool to solve this.

The Tech Stack
I wanted to keep it free and fast, so I used:

Python (Streamlit): For the frontend UI.

Supabase: To store the database of captions.

Hugging Face: To host the application.

How it works
I indexed the captions from the last 12 years of the channel. The search engine allows you to type any phrase, and it returns the exact YouTube video timestamped to that second.

You can try the live demo here: ThingsMrBeastSaid.com

The Challenge: Speed
The biggest hurdle was making the search instant. Searching through millions of lines of text can be slow. I used Supabase's ilike filters to match keywords against the caption database, which returns results in under 200ms.

I also implemented "Deep Linking," so you can share specific search results. For example, here is every time he mentions his Private Island or his Lamborghini giveaways.

Future Plans
I am currently logging search queries to see what people are looking for most. Surprisingly, "Money" isn't the top result—people are mostly looking for specific funny moments from the crew (Karl, Chandler, etc.).

If you're a developer or just a fan, check it out and let me know what you think!

Live Link: www.thingsmrbeastsaid.com

Top comments (0)