A CV that screening software can read
Next to the designed CV there is an ATS design, one column with no tables and plain headings, and both PDFs carry a title, author, subject and skill keywords in their metadata.
The problem
Many companies pass CVs through an applicant tracking system or an AI screener before a person sees them. These tools read the PDF's text in order. Tables, two-column layouts and text in headers or footers come out mixed up or are lost, so a good CV can score badly.
How dok.sh solves it
- Two designs from the same content. Classic is the designed layout. ATS is one column in reading order, with no tables, standard section headings (Summary, Experience, Skills, Education), the contact line as plain text and labelled links ("LinkedIn: …").
- Positions in the order parsers expect: "Title | Company", then "Dates | Location", then bullets.
- PDF metadata for both designs: title ("Name – headline"), author, subject and up to 30 skill keywords.
- The admin sets which design the site uses by default;
?template=ATSon a download link always gives the ATS version, and its file name ends in-ats.

The result
A recruiter gets a designed CV, a screening tool gets clean text, and both come from the same data.
Technologies
PDF, ATS, Java
- Published: 2026-11-03
- Project: dok.sh — Portfolio and CV Platform
- Case study