Frontier Lab Hiring Documentation

OverviewMethodologyFAQRecordsDownloads
Show sidebarRecords

Records

The Frontier Lab Hiring dataset is shipped as six CSV files: the daily count of open roles by category, the daily count of open roles by country, the daily count of open roles by city, one row per posting, the latest text of each posting, and one row per revision of a posting’s text. The posting column identifies the same role across the postings and posting-texts files, and posting_key in the revisions file joins to it.

We provide a guide to each file’s fields below, with example values for reference.

Open roles by category

frontier_lab_hiring_open_roles_by_category.csv — the daily series behind the chart: one row per scrape date, lab and sub-category, counting that lab’s postings in that sub-category open on that day. Zero counts are included. Summing rows over subcategory within a category gives the category series.

ColumnTypeDefinitionExample value

Open roles by country

frontier_lab_hiring_open_roles_by_country.csv — the daily series behind the chart’s country view: one row per scrape date, lab and country, counting the postings open on that day whose location lists that country. A posting is counted once per country it lists, so the rows for a day add up to more than the lab’s open postings. Zero counts are omitted. Rows with lab set to “All labs” give the two labs’ counts added together, and the country “Unmapped” collects locations that could not be mapped to a country.

ColumnTypeDefinitionExample value

Open roles by city

frontier_lab_hiring_open_roles_by_city.csv — the country series broken down to cities: one row per scrape date, lab, country and city, counting the postings open on that day whose location lists that city. A posting is counted once per city it lists. The city is Remote for a listing that is remote within the country and Unspecified for one that names a country or region but no city. Zero counts are omitted, and rows with lab set to “All labs” give the two labs’ counts added together.

ColumnTypeDefinitionExample value

Postings

frontier_lab_hiring_postings.csv — one row per posting ever observed: its lifetime in the archive, its latest label, and its location. The countries column lists the countries its location maps to, cities lists Country: City pairs (with Remote in place of a city for a listing that is remote within the country, and Unspecified for one that names a country or region but no city), and location_source records whether the location came from the board’s location field or was inferred from the advert’s title or text.

ColumnTypeDefinitionExample value

Posting texts

frontier_lab_hiring_posting_texts.csv — one row per posting: the advert’s text as last observed. On its own this table is served gzipped, as frontier_lab_hiring_posting_texts.csv.gz; the ZIP contains the plain CSV.

ColumnTypeDefinitionExample value

Revisions

frontier_lab_hiring_revisions.csv — one row per version of a posting’s text, with the label given to that version. A posting has one revision when it first appears and gains another each time the lab changes its title, department or advert text; the daily counts use the label in force on each day.

ColumnTypeDefinitionExample value