My web site is one of my favorite yaks. There are many like it, but it is mine.
I build it as a static site published with Netlify. I track the content & compiler source in Git on GitHub, and use Netlify's build farm for recompilation. A Zapier calendar hook triggers a nightly rebuild so external content (such as citation counts) refresh regularly.
- Gulp ties everything together
- Remark for most of the content processing
- Pandoc for the few pages that require more complex formatting than Remark can handle
- Rehype for reprocessing HTML
- Nunjucks for layouts
- Hastscript for generating little bits of HTML in the compilation process
- PostCSS to make CSS more usable
- Lunr to implement site search
- KaTeX for math rendering
- iText pdfHTML for HTML-to-PDF
One of the particularly cool features of this setup is my publication pages: each publication has a page written in Markdown; embedded HTML at the top of the page serves as single-source-of-truth for my publications across the whole site. My CV is updated from this data, as is my publication list and other places where publications show up (project pages, blog posts, etc.).
I source citation statistics from Microsoft Academic via Cognitive Services Labs, updated every time the site rebuilds.
I use iText pdfHTML to render my CV into a consistent and downloadable PDF. The HTML CV itself prints well (the print CSS is how I generate the PDF), but the pre-rendered PDF provides niceties like good headers and footers. My build process includes a small Java program that calls the relevant iText APIs to convert generate the PDF after the site's HTML content is compiled.
Icons are from the excellent Noun Project.
I don't make the Git repository and history of my web site public, but am happy to share upon request with people interested in seeing the code.