I created a makefile that takes the markdown and generates pdf and other outputs using pandoc with reveal-js. It works great for me except for the part about managing complex figures that I need a gui to draw. For figures that I create in python, that's great (I believe quarto was mentioned earlier, and that works great for more dynamic figure too). But if I'm annotating figures, opening up ppt and adding an arrow and text box is so easy.

I try to model my figures after these suggestions (it means it's not easy to make figures without a visual editor): https://www.youtube.com/watch?v=Unzc731iCUY

Yes, reveal-js is missing an easy to se wysiwyg editor, I agree.