Skip to content

What 985 companies are actually hiring for

A census of 148,071 open roles across 985 company career pages. Software engineers lead, and sales outnumbers every data and AI role combined.

Jobzyl5 min read

Figures on this page are current as of

Across 985 company career pages we pull in full, software engineer is the most advertised role by a wide margin: 9,837 open postings, almost three times the next role. The finding that did not fit the narrative is underneath it. These companies are advertising 3,700 account executive roles against 2,159 for data scientist, AI engineer and machine learning engineer combined. Sales outnumbers every data and AI title on the list put together, and it does so across 333 separate companies.

148,071

open postings, across 985 company career pages

Deduplicated on the same key the product uses, so a role posted to several boards counts once. This is a count of postings, not of rows in our index, and the two are not the same number.

Data as of

Why this reads company career pages and not the whole index

Most of what we ingest is keyword-driven. Our scraper asks each board for a list of terms across a list of countries, so the mix of roles in those rows is a picture of what we searched for. Ranking roles across the whole index would be ranking our own configuration file, and it would be wrong in a way nobody could see from the outside.

Four sources work differently. Greenhouse, Lever, Ashby and SmartRecruiters are pulled a company at a time, whole board, no query. Within that set of companies the role mix is a census, and nothing we chose to look for shaped it.

That buys accuracy at the cost of reach. These are companies that publish on four modern applicant tracking systems, which skews technology-heavy and away from the public sector and small employers. Every number here is a claim about these 985 companies. None of it is a claim about the labour market.

The ranking

Open postings by role, 985 company career pagesTitles are matched by pattern, so one posting can count under more than one role and the bars do not sum to the total.
Show the numbers
Open postings by role, 985 company career pages
CategoryValue
Software engineer9837
Account executive3700
Product manager2783
Designer2335
Project manager2009
Marketing manager1249
Data engineer1111
Customer success manager1022
Security engineer912
Data scientist832
Recruiter737
AI engineer698
Machine learning engineer629
Accountant541
DevOps engineer341

Data as of

Software engineer takes 9,837 postings across 582 companies. Below it the list is less engineering than the platforms' reputation suggests: account executive, product manager, designer and project manager occupy four of the next five places, and together they outweigh software engineer.

Designer at 2,335 postings is the one that surprised me most in ordinary use. It sits above data engineer, data scientist, AI engineer and machine learning engineer combined, across 395 companies. Design is usually reported as the first function cut when hiring tightens.

The data and AI roles are smaller than they sound

RolePostingsCompanies
Data engineer1,111277
Data scientist832239
AI engineer698184
Machine learning engineer629155
DevOps engineer341147

Two things worth pulling out. AI engineer has overtaken machine learning engineer as a title, 698 to 629, which is a naming shift rather than a capability one and is the kind of thing that only shows up in postings. And data engineer outranks data scientist, 1,111 to 832, which has been true in practice for a while but rarely gets stated with a count next to it.

I also checked a claim I expected to make and could not. On a looser title match, machine learning appeared to outrank data scientist. Tightening both patterns to the same strictness reversed it. The looser number was measuring pattern breadth, not hiring, so it is not reported above.

Where these roles are

Where the census sits, by countryAn ATS board is pulled with no query location, so the country is derived from each posting's own text.
Show the numbers
Where the census sits, by country
CategoryValue
United States59346
Not resolved16051
United Kingdom9300
France7084
India7029
Germany5424
Canada5179
Brazil2965
Spain2909
Remote2402

Data as of

The United States takes 40.1% of the census. The second largest category is not a country: 10.8% of postings do not carry enough location text to place at all, and that is reported rather than quietly dropped, because dropping it would inflate every country share below it.

The United Kingdom is 6.3%, France 4.8%, India 4.7%, Germany 3.7% and Canada 3.5%. Only 1.6% is explicitly labelled remote, which is a fact about how these companies write location fields rather than about how often the work is remote.

What has not moved, and why we are not telling you

The obvious second half of a report like this is what changed since last month. We are not publishing that, because nothing we have measures it.

Our index holds six weeks of ingest. The largest source in it arrived two weeks in, on 17 August, carrying 40% of all rows. Among the six sources that do span the whole window, weekly volume swings 8.4 times between its low and its high, from 108,990 rows to 910,338, and that pattern is our own sweep schedule rather than employers. Inside the career-page census the first week is a seeding backfill of every board at once, and the most recent week's rise tracks discovery adding 232 new companies to the set.

A new source going live looks exactly like a hiring surge. So there is no window here in which a week-over-week change measures demand, and every riser we could have reported would have been our own crawler.

This post is the baseline instead. The queries, the windows, the title patterns and the row floor are all declared in the post's own configuration, so the next edition runs the same cuts against a comparable window, and movement becomes reportable the moment there are two snapshots taken the same way.

Methodology

Source. Four applicant tracking platforms, ingested a company at a time with no search query: Greenhouse, Lever, Ashby and SmartRecruiters. Keyword driven sources are excluded, because their composition reflects our keyword list.

Data pulled 13 September 2026. Ingest window 10 August to 13 September 2026.

Deduplication. Collapsed on the product's own dedupe key before counting, so the same role advertised on several boards counts once. Rows in the index run about 1.34 per distinct posting; every figure here is a posting count.

Counts. 148,071 distinct postings across 985 distinct companies. Role matching is by title pattern, so a posting can count under more than one role and role counts do not sum to the census. Any role or country below the row floor is suppressed rather than printed.

Known limitations. Which companies are in the set is decided by ATS discovery, which is not random and skews technology-heavy. 10.8% of postings cannot be placed to a country. No liveness check has been run on this subset, so a posting counted here may already be closed. No movement is reported, for the coverage reasons set out above.

Every figure here comes from the jobs Jobzyl has indexed, and carries the date it was pulled. Subscribe by RSS, or run the search yourself.