Here's the video - https://www.youtube.com/watch?v=x78G7IQzBc4

TLDR: I walk through a way of building a self-hosted recommendation system. I'm writing a script to track my own behavior with the metrics I actually want vs whatever YouTube is doing. Storing it locally so I can easily manipulate it when I want to (remove or add topics over time). And fetching videos my scraping Youtube search with a search query from the topics I'm storing over time - END TLDR

This is a derivative of a project I released last year called myAlgorithm where I made this generically for any sort of website. You can check code out on my Github (https://github.com/jawerty/myAlgorithm)

Overall, these past few weeks I've been producing content around web scraping, fine tuning LLMs, using word2vec on your own scraped dataset and more. I want more engineers to learn how to source their own data and build AI tools on top of it without the need for endless APIs.

If you have any questions about this stream or any other streams I've done lmk and there's plenty more to come!