Skip to content

Resume parsing & matching

See the evidence behind every candidate-job match.

HireIQ turns resumes into structured, source-linked evidence and compares it with the requirements of each role. Reviewers see which requirements the available evidence supports, which remain unknown, and the exact text that supports each conclusion.

Create a workspaceTalk to us

hireiq / candidates / ananya-rao / match

Sample data

Ananya Rao

Senior Backend Engineer · Applied via careers page

Expert Review

Requirement coverage

  • Go in production services· Required

    Supported

    “Rewrote the payments ledger service in Go and moved it to production.” — Resume · page 1

  • Event streaming with Kafka· Required

    Partially supported

    “Built Kafka consumers for order events.” — Resume · page 2

  • PostgreSQL performance tuning· Preferred

    Unknown

    No evidence in the submitted documents yet.

Next step decided by a reviewerAdvance or reject with a reason
Illustration of requirement-level matching with quoted resume evidence and explicit unknowns.

The problem

A match percentage is not an explanation.

When hundreds of applications arrive for one role, reviewers need to know why a candidate surfaced, not just that a model scored them highly. HireIQ keeps the original document, the extracted evidence, and the job requirement separate, so every match can be traced back to the words a candidate actually wrote.

How it works

The workflow, step by step.

  1. 1

    Ingest the original document

    PDF, DOCX, and plain-text resumes up to 10 MB are stored as submitted. Scanned PDFs are read with OCR in English and Hindi before extraction.

  2. 2

    Extract structured evidence

    Skills, experience, and education are captured with the quoted source span, so a claim never exists without the text it came from.

  3. 3

    Understand the role

    Job descriptions are converted into structured requirements that separate hard constraints, required capabilities, and preferences.

  4. 4

    Compare requirement by requirement

    Each requirement is marked Supported, Partially supported, Unknown, Contradicted, or Unsupported, with the evidence that led to that state.

Capabilities

What’s included in resume parsing & matching.

  • Source-linked evidence

    Evidence records keep their source, a content hash, and extraction metadata. New information creates a new record instead of overwriting history.

  • Unknown is not a no

    Missing evidence is shown as Unknown rather than treated as a failure, so strong candidates with sparse resumes are not silently excluded.

  • Skill aliases without false equivalence

    A curated skill ontology recognizes aliases and abbreviations while keeping related technologies distinct. React is not counted as React Native.

  • Duplicate and identity review

    Candidates who apply through several channels are linked through explicit identity signals, with merge decisions recorded rather than inferred silently.

  • Talent rediscovery

    Past applicants can be matched to new roles and invited back, excluding anyone who withdrew, opted out, or requested deletion.

  • Human relevance judgments

    Reviewers can record whether a surfaced candidate was relevant, building a labelled record that is used to evaluate matching quality.

Controls and safeguards

  • Candidates who opt out of AI processing are skipped by AI-assisted analysis.
  • Published job specifications are versioned, so past matches remain reproducible.
  • Coverage ordering is an aid to review, never a hiring decision.
  • The original document is kept, so extraction can be re-run when parsing improves.

Who it helps

Recruiters
Work through large applicant pools by requirement coverage instead of keyword hits.
Hiring managers
Read the quoted evidence behind a shortlist before meeting anyone.
Administrators
Audit how a match was produced and which specification version it used.

Works together with

FAQ

Resume parsing & matching: common questions

Which resume formats are supported?

Hosted applications accept PDF, DOCX, and TXT files up to 10 MB. Scanned PDFs are processed with OCR. Spreadsheet imports accept CSV and XLSX files.

Does HireIQ give each candidate a single score?

No. Matching reports a state for each job requirement with supporting evidence. Ordering by evidence coverage helps reviewers prioritize, but there is no universal candidate-quality score.

What happens when parsing fails?

The application keeps the original file and shows the parsing status, so an administrator can retry extraction or replace the document. Nothing is inferred from a file that could not be read.

Bring structure to your next requirement.

Create a workspace, post your first role, and see every applicant’s evidence in one pipeline.

Create a workspaceTalk to us