| configs |
|
||||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| license | bigscience-openrail-m |
Daily snapshots of data science / machine learning job postings scraped from LinkedIn, one parquet file per scrape run. All splits share one schema (2023 → today); historical splits were migrated in 2026-07 — the original 8-column data is preserved at revision v1-schema.
The same job_id recurs across splits (a posting stays live for days) — that's the longitudinal signal. For a unique-jobs view, dedup on job_id keeping the row with max scrape_dt.
Publicly visible posting fields only: title, company, location, description, salary where posted. Fields that are login-gated or Premium-gated on LinkedIn (applicant statistics, company insights, hiring-team contacts) are never published. Missing values are NULL, not sentinel strings; fields not observable historically are NULL in pre-2026-07 splits.
BigScience OpenRAIL-M — research and educational purposes only. Commercial use and redistribution are prohibited.
Generated by linkedin-jobs-scraper (Python + Playwright).