I've encountered a real rabbit hole here, and I'm not sure what to do about it. The naive view is that LaTeX does typesetting the best way. If you want better typesetting, use LaTeX, and if you're using LaTeX, you're done.
This turns out not to be true, which is upsetting to those of us ingrained in the LaTeX ecosystem and used to thinking of it as the best way. The first thing you're likely to learn is that you need the microtype package. Without this package, kerning is not as good as it could be. (Which raises the question - if LaTeX is the final form of a typesetting language, why not do it right by default?)
But the problems don't end there. Modern OTF / TTF fonts have a lot of typesetting information that isn't fully retained if you convert the fonts to use with LaTeX. So if you want modern font support, you can't use LaTeX at all, you switch to XeTeX!
At this point you realize that the microtype package you previously learned was crucial isn't compatible with XeTeX, so you can't use it.
So it's a Catch-22. LaTeX is for people who are obsessed with perfect typesetting. But perfect typesetting can't be achieved (currently, to the best of my knowledge) in any TeX family language.