A Markdown résumé,
as a clean PDF

A résumé is mostly a short list of headings, dates and bullets, and almost all the time spent on one in a word processor goes into fighting the layout rather than improving the content. In Markdown the structure is unambiguous, the file is small enough to keep in version control, and tailoring it for a specific role is a text edit rather than a formatting exercise.

Most large employers parse your PDF before a person sees it, and most parsing failures come down to the file not containing readable text in a sensible order. A few choices matter more than the rest.

• The text must be real text, not a picture of text — exporting a design tool's canvas as an image is the most common way a résumé becomes unreadable, and does not happen here
• Use headings for sections and bullets for achievements, so the reading order matches how a human would read it
• Avoid multi-column layouts and avoid tables used purely for positioning; parsers read them in an order you did not intend
• Keep contact details in the body rather than in a header or footer, which some parsers skip
• Links to LinkedIn or a portfolio stay clickable, so write them as [text](url) rather than pasting a bare URL
• Fonts are embedded, so the document looks the same on the reviewer's machine as it does on yours

The style panel is the fastest way to gain or lose a page without cutting content you wanted to keep. Reduce the margins, drop the body size by a point, and choose a more compact typeface — Inter and Source Serif set tighter than Merriweather at the same size. Turn page numbers off, since a one-page résumé does not need one, and leave the title page off so your name stays at the top of the first page.

Name as the top-level heading, contact details on the line beneath, then one second-level heading per section — Experience, Education, Skills. Under each role, a third-level heading with the job title and dates, then bullets for what you actually did. Nothing more elaborate is needed, and anything more elaborate is where parsers start guessing.