I rarely look forward to updating my résumé, let alone doing it twice. As a hiring manager, though, I’ve appreciated candidates who share both a PDF and a web résumé, especially when the website adds more context.

Keeping two versions aligned

My résumé is available as a page on this website and a downloadable PDF. A change to a job title, employment date, or description of my work needs to appear in both. Editing the documents separately would mean making each change twice.

I keep the shared content in resume.yml. The website and PDF each read this file and apply their own layout. I edit a role’s description once, then rebuild both outputs.

What lives in the YAML file

YAML is a plain-text format for structured data. Named fields hold values, indentation groups related fields, and a dash starts a list item.

My file follows the input structure of RenderCV, a tool that generates résumé documents from YAML. Here is an excerpt with one role and one highlight:

cv:
  name: Antoine Ménard
  sections:
    Experience:
      - company: Snap Inc.
        position: Software Engineering Manager
        start_date: "2018-02"
        end_date: "2024-06"
        location: Los Angeles, CA
        highlights:
          - Contributed to Camera initiatives including Director Mode
            and Home & Lock Screen Widgets, supporting implementation,
            reliability, and post-launch iterations.

The cv field holds the résumé content. Under sections, Experience is a list of roles, each with an employer, title, dates, location, and highlights. The wrapped lines in this example form one highlight. Quoting the dates keeps them as text.

The full file also contains profile links, a summary, education, patents, and languages, plus fields used only by the website. PDF design and language settings live under design and locale.

This is my “one source of truth” for résumé content. Each output decides which fields to display and how to present them.

Two paths from the same source

Astro builds the web page from data and HTML templates. RenderCV uses the Typst typesetting tool to lay out the PDF’s pages. The two paths branch from the same file:

One editable source resume.yml
Website
  1. Web loaderDates, links, sections
  2. AstroHTML template
  3. Web résumé/resume/
PDF
  1. PreparationGenerated YAML
  2. RenderCVThen Typst
  3. PDF résuméantoine_menard_resume.pdf
One source file feeds two independent rendering paths.

The TypeScript loader in src/lib/resume.ts reads the file with the yaml package, constructs profile links, formats dates, and collects the supported sections. The example’s dates become Feb 2018 – Jun 2024.

src/pages/resume.astro turns that data into headings, lists, and role descriptions. Astro renders the template as HTML during the site build.

The loader recognizes specific section names. A new section, such as certifications, needs support in both the loader and Astro template to appear on the website.

The PDF path starts with tools/resume-pdf/prepare-input.mjs. It reads resume.yml, checks attachment fields with the résumé loader, then writes a generated copy to .rendercv/resume.yml with each experience entry’s attachments removed.

Attachments are links to supporting material, stored with a title, URL, and optional publisher label and icon. These nested records cannot pass validation in the repository’s RenderCV 2.8 setup, so the preparation step removes them from the PDF input. This adaptation is specific to that attachment structure; RenderCV supports custom entry keys.

The generated copy keeps the rest of the document, including design and locale. The script also copies the tracked header template. The editable resume.yml stays untouched.

1. Original source resume.yml
  • Résumé content
  • Experience attachments
  • design and locale
Stays unchanged
2. Generated copy .rendercv/resume.yml
  • Résumé contentKept
  • Experience attachmentsRemoved
  • design and localeKept
Passed to RenderCV
Preparation removes experience attachments from the copy. The website still reads them from the original.

RenderCV turns the generated YAML into a Typst document using the selected theme. Typst produces the PDF; the website’s HTML and print styles are never involved.

The generated YAML and PDF are ignored by Git and replaced during PDF generation. Lasting edits belong in the original YAML or presentation code.

Shared content, different presentations

The web résumé has an experience timeline, role icons, and attachment cards linking to supporting material. It follows the site’s light and dark themes and reflows to fit a narrow screen.

The PDF uses RenderCV’s engineeringresumes theme. The YAML’s design settings control the typography, and a header template stored in the repository customizes the profile-link labels.

A highlight can use identical wording in both formats while occupying different numbers of lines.

Sharing the content avoids editing the same facts twice. The two layouts still need to be maintained and checked separately.

After editing resume.yml, regenerate the PDF with npm run resume:pdf, then build the site with npm run build. An ordinary build can otherwise reuse an older PDF. The Cloudflare build path uses CF_PAGES=1 to regenerate the PDF first; npm run build:cloudflare reproduces it locally.

Preview the result and find the changed text in both /resume/ and its linked PDF. Check every PDF page for clipping and awkward breaks, and inspect the web page at a narrow width. A successful build or download alone does not prove both outputs are current.

To adapt this pattern, start with the data structure your PDF tool accepts and map it into the web layout. Keep web-only additions in the source, then filter them from the generated PDF input as needed.

And LinkedIn?

For my LinkedIn profile, I’m keeping things ad hoc for now: I ask my current AI client to update it from the latest resume.yml. I might later turn that into a skill. It would provide reusable instructions for updating LinkedIn alongside my résumé.