Skip to content

Dataset · dtunkelang

348K job vectors and metadata for search

dtunkelang/jobs-demo

Texts, JSONL metadata, and ready-to-use embeddings for job postings and search queries.

What's inside

Search engine artifacts contain titles, JSONL metadata, and pre-computed embeddings (bge-small and te3-large) for 347,900 job postings from four sources: Open-Apply, LinkedIn, JobStreet, and USAJobs. The package also includes a cache of pre-computed vectors for approximately 196,000 popular search queries.

What you can do with it

  • Testing and configuring hybrid and semantic search (RAG) for jobs
  • Benchmarking vector search speed and accuracy without generating embeddings
Size
up to 1K rows
Downloads this month
42.7K
Likes
2

How to load it

from datasets import load_dataset

ds = load_dataset("dtunkelang/jobs-demo")