How CV parsing works (and how to fix parse errors)

What the parser extracts

CVViZ uses two parsers — a legacy one and an OpenAI-backed parser (AdvanceParserNew) when your workspace has it enabled. The advanced parser's UI says: "Our AI extracts contact details, work history, and skills with high accuracy".

Concretely, the parser pulls:

• Name — first and last.
• Email.
• Phone (Contact Number).
• Work history — past job titles, employers, descriptions.
• Education (qualifications).
• Skills — surfaced as topskills on the candidate profile.
• Years of experience.
• Current company and job title.
• Social links — LinkedIn, GitHub, Twitter, Instagram, Facebook, Behance.

The original CV file is also stored — accessible from the candidate's Resume tab.

Reviewing parsed data

After upload, click into the candidate. The Profile tab shows parsed fields; the Resume tab shows the original document. Edit any parsed field inline if it looks wrong.

Re-parsing

If a CV parsed badly, you can re-parse it.

• Bulk — select multiple candidates from a list and click Re-Parse Resumes (with a sync icon) in the bulk actions bar. CVViZ asks for confirmation: "Are you sure you want to re-parse selected resumes?"
• Single candidate — re-parse from the candidate's actions menu when needed.

Re-parsing re-runs the AI extraction and updates the profile fields. Manual edits you made before re-parsing may be overwritten — make sure that's what you want.

Tips for cleaner parsing

• Selectable text PDFs only. Image-only scans (typed photos, screenshots) fail. Run them through OCR first if you can.
• Standard headings help. Sections labelled "Experience", "Education", "Skills" parse cleanly.
• One-column layouts keep work-history order intact.
• Avoid creative templates — sidebars, mixed icons, and design-heavy layouts confuse the parser.
• Common file formats — PDF and DOCX work best.

Replacing the CV file

Got a newer version of the candidate's CV? Upload it from the candidate's Resume tab. The new file is stored alongside the previous version.

What to fix manually vs re-parse

• If 1–2 fields are off, fix them inline. Faster than re-parsing.
• If most fields look wrong, re-parse first — sometimes a re-run resolves it.
• If even re-parsing produces poor results, the source CV is probably the culprit (image-only or unusual layout).

Advanced vs legacy parser

If your workspace has theflag on, the new AI-backed parser (AdvanceParserNew) handles uploads. Otherwise, the legacy parser (AdvanceParse) is used. The advanced parser is more tolerant of unusual layouts and tends to extract skills more accurately.