It sounds like (from the shout-out in the post) they're using https://math.nist.gov/~BMiller/LaTeXML/ to convert the paper's LaTeX into HTML, not from PDF.

The most versatile tool I know of for converting various document formats, including PDF to HTML, is the oss ebook tool Calibre: https://manual.calibre-ebook.com/conversion.html

I have seen https://pdfbox.apache.org/ used for extracting text from PDFs for analysis, but you won't get HTML output.