Parse + Extract

Both artifacts in one call, delivered separately: a parse object (the document) and an extract object (the 24-field profile) — persist each on its own.

One request, two independent artifacts — published as an explicit endpoint so your backend stays a simple pair of calls: fire /v1/autofill and /v1/parse-extract in parallel on upload, persist both artifacts from this response, and merge on submit (the candidate's own form edits win over the extracted fields).

Request

Send multipart/form-data — same contract as /v1/parse: file or source_url, max 20 MB, first 10 pages.

Response 200

Two top-level objects, not nested in each other: parse (exactly what /v1/parse returns) and extract (the profile):

{
  "parse": {
    "markdown": "# Jane Doe\n...",
    "chunks": [ ... ], "splits": [ ... ],
    "metadata": { "filename": "resume.pdf", "duration_ms": 16408 },
    "instant_fields": { ... }
  },
  "extract": {
    "candidate_data": {
      "first_name": "Jane", "last_name": "Doe",
      "work_experience_details": [ ... ], "resume_skills": [ ... ]
    },
    "metadata": { "duration_ms": 4200 }
  }
}

Store parse and extract as two separate records; the raw markdown is for your records — never render it to the candidate. All 24 fields: Response schema.

Timing

Typically 10–25 seconds total for a typical CV (parse + extraction in sequence, one round-trip).

Example

curl -X POST "https://prs.recruspace.com/v1/parse-extract" \
  -H "X-API-Key: $KEY" -F "file=@resume.pdf"