NextGig-Rocks/global-job-postings-multi-ats · Datasets

NextGig-Rocks/global-job-postings-multi-ats · Datasets at Hugging Face Multi-ATS Job Postings Snapshot (June 2026) A dated snapshot of 112,816 job postings collected from 23 different Applicant Tracking Systems (Workday, Greenhouse, Lever, Ashby, iCIMS, SmartRecruiters, Oracle Cloud, Workable, Paylocity, and more) and parsed into a clean, consistent 47-column schema with a large language model. Unlike typical single-source job dumps, every row is enriched with extracted skills, salary ranges, qualifications, experience level, work model, and geocoded location (lat/lon) — useful for salary modeling, skills-demand analysis, NLP, and labor-market research. Data sourced and structured by NextGig— for live, current listings visit https://www.nextgig.rocks. Highlights - 112,816 postings · 13,412 companies · 23 ATS platforms · ~540 geo regions - 47 structured columns - ~51,800 postings (46%) include structured salary min/max - Skills extracted for ~71%; geocoded lat/lon for ~94% - LLM-generated role summaries (98%) Quick start ``` from datasets import load_dataset ds = load_dataset("NextGig-Rocks/global-job-postings-multi-ats", split="train") print(ds0) ``` List-type columns (`skills_required`,`minimum_qualifications`, etc.) are stored as JSON strings — parse with`json.loads`. What this is NOT A point-in-time research snapshot, not a live feed. It deliberately contains no application URLs, no apply links, and no source identifiers — it cannot be used to apply to jobs or to power a live job board. Known limitations - `date_posted` is sparse (~20%); not suitable for time-series/trend analysis. - `country`/`city` are derived from geocoded strings and are approximate. - The`job_description` field is an LLM-generated 2–4 sentence factual summary of the role (not the original posting text) and may contain occasional errors. - Some fields (e.g.`years_experience_numeric`,`equity`) are sparsely populated. License CC BY 4.0 — free for commercial an