rev.2026.04
§ dataset · schema
Every field, and how often it’s filled.
Two tables.
jobs holds one row per posting; job_applicant_history holds a timestamped applicant count per observation. Coverage is measured on the live archive, and example values are copied from the free sample — nothing here is invented, and nothing is listed that the scraper does not populate.§ 01 · jobs · 18 field groups
The jobs table.
| field | type | what it is | example | filled |
|---|---|---|---|---|
| applicant_count | integer | Applicants LinkedIn reported at last observation. Almost nobody else sells this. | 117 | 8% |
| applicant_count_is_capped | boolean | True when LinkedIn showed a ceiling (“over 200”) instead of a number, so a flat series isn't mistaken for stalled interest | false | 100% |
| salary_min · salary_max | numeric | Parsed out of the posting's own salary text | 40000 · 48000 | 10% |
| salary_currency · salary_interval | text | Currency code and period. 40+ currencies present, not USD-only | EUR · yr | 10% |
| salary_range | text | The raw salary string, kept so you can re-parse if our regex missed a locale | ₹1,000,000.00/yr - ₹3,000,000.00/yr | 10% |
| linkedin_job_id | text · primary key | LinkedIn's own posting id. Stable, joinable, dedupes across regions | 4459780495 | 100% |
| title | text | Role title exactly as posted | Java Fullstack Developer | 100% |
| company · company_slug · company_id | text · text · integer | Company name, LinkedIn slug, and numeric company id for joins | Viraaj HR Solutions · viraajhrsolutions · 30211195 | 99% |
| location · region | text | Location as posted, plus the region partition it was found in. 218 regions worldwide | Mumbai Metropolitan Region · Maharashtra, India | 100% |
| description | text | Full job body, plain text. Person-shaped contact details removed | About The Opportunity… | 100% |
| level | text | LinkedIn's seniority label. Note: ~2 in 3 postings carry “Not Applicable” — that is LinkedIn's data, not a gap in ours | Mid-Senior level | 100% |
| employment_type | text | Full-time · Part-time · Contract · Internship · Temporary | Full-time | 100% |
| job_function · industries | text | LinkedIn's function and industry labels | Engineering and IT · Software Development | 99% |
| industry_ids · title_id | integer[] · integer | LinkedIn's numeric ontology ids — the stable join keys behind the text labels | [4] · 10738 | 99% |
| posted_date · posted_at_precise | timestamptz | When LinkedIn says it was posted. The precise variant is to the minute | 2026-08-31T08:45:35Z | 96% |
| first_seen_at · last_seen_at | timestamptz | When we first indexed it and last re-saw it. The two together are the posting's observed lifetime | 2026-08-31T09:20:02Z | 100% |
| is_active | boolean | Flips false when a posting stops being re-seen. Time-to-fill falls out of first_seen_at → this flipping | true | 100% |
| linkedin_url · canonical_url | text | Back to the posting itself | linkedin.com/jobs/view/4459780495 | 100% |
§ 02 · job_applicant_history
The applicant-count time series.
Captured by a database trigger each time a re-crawl sees a different applicant count for a posting. Join on linkedin_job_id and order by observed_at to get the series; the slope is the posting’s applicant velocity.
| field | type | what it is |
|---|---|---|
| linkedin_job_id | text | Foreign key to jobs.linkedin_job_id. |
| applicant_count | integer | The count LinkedIn displayed at this observation. |
| is_capped | boolean | True when LinkedIn showed a ceiling (“over 200”) rather than a number. |
| observed_at | timestamptz | When the observation was recorded. One row per re-crawl that saw a change. |
The same columns come back from the free sample, the snapshot, live access, and the jobs API. Field names never change between them.