Covers Quantum Field Theory and General relativity in as approachable a manner as I imagine is possible while still going deep into detail: https://www.youtube.com/c/viascience/videos
Theoretical machine learning, Information theory and probabilistic inference: https://www.youtube.com/user/mathematicalmonk/videos
Heavy emphasis on analysis and measure theory: https://www.youtube.com/c/brightsideofmaths/videos
While ScienceClic's coverage of differential geometry topics is mostly a thin vertical slice of the subject, it covers it with incredible visualizations and pedagogical approach: https://www.youtube.com/watch?v=xodtfM1r9FA&list=PLu7cY2CPiR....
https://youtu.be/5ESJH1NLMLs "Children of the Magenta Line" by Warren Vanderburgh https://youtu.be/YvEB05xdAy4 "Total Synthesis of Vitamin B12" by RB Woodward
You may tell me I'm cheating by posting videos that are obviously not "native" to YouTube, but YouTube was the way I found them, how I watch them, and how I share them with other people.
OK, so YouTube-native videos are hard mode. Try https://youtu.be/WHASYE2e5Xo I suppose? I like it. It has life lessons I'd argue are at least as valuable as the videos you're thinking about.
Kling is building an operating system and a web browser, and frequently relases videos about the work. Most go through development of a single feature. The videos are unedited to show all of the development steps: design, implementation and debugging. A Masters program usually has more theory and larger scope, but this is more hands-on and in-depth.
Math and art are thing I like, just keep hard blocking stuff you don't like and upvote stuff you do.
Other than occasional captured lectures, everything 'educational' on YouTube seems to me to be optimising for that TED-style sense of wonder / happy familiarisation with the content, which is poorly associated with learning outcomes.