Developer guides

Workday career site JSON: the jobs endpoint, facets, the 2,000-row cap and big boards

Every Workday career site loads its jobs from a JSON endpoint, POST https://{tenant}.{wdN}.myworkdayjobs.com/wday/cxs/{tenant}/{site}/jobs, which returns 20 rows per call and silently wraps back to page one after 2,000 rows. Source: Arcadium Data, data as of 2026-10-25.

Workday is the most common system among large employers: it is behind the career sites of banks, chipmakers, retailers and hospitals. It has no published public jobs API, but the sites you see at company.wd5.myworkdayjobs.com are single-page apps that call the same JSON service you can call. That service is stable enough to build on, and has exactly the quirks that make Workday the costliest ATS to crawl.

Among the boards we crawl, Workday has around 3,800 employers (17% of the live directory) and takes most of the crawl cost. This guide covers the request, facets, the cap and a way around it, the detail call, and what to do about pace.

The official route: the career site's own JSON service

Find the three parts of the address

A board needs a tenant (the company's id on Workday), a data centre (wd1, wd3, wd5, wd12 and others) and a site (the career site name; one tenant can have several, such as an external site and an internal one). All three are in the career-site URL: https://nvidia.wd5.myworkdayjobs.com/en-US/NVIDIAExternalCareerSite has tenant nvidia, data centre wd5 and site NVIDIAExternalCareerSite; the en-US is a language and is not part of the address. You cannot guess these from a company name. Find them in the careers page of the company, or in lists of ATS board URLs (see companies using an ATS).

The list call

curl

curl -s -X POST "https://nvidia.wd5.myworkdayjobs.com/wday/cxs/nvidia/NVIDIAExternalCareerSite/jobs" \
  -H "content-type: application/json" -H "accept: application/json" \
  -d '{"appliedFacets":{},"limit":20,"offset":0,"searchText":""}'

The answer has total, jobPostings (20 rows with title, externalPath, locationsText, postedOn and bulletFields, which usually holds the requisition id) and facets. The page size is 20; larger limits are refused or clipped. postedOn is relative ("Posted 3 Days Ago", "Posted 30+ Days Ago"), so the list gives day-level dates at best.

The cap and how to get past it

One query returns at most 2,000 rows. Past that, paging does not end: the list silently restarts at page one, and a naive loop collects the same 20 jobs forever or stops with a wrong total. Detect it by stopping when a page contains only ids you have already seen. To read a board of more than 2,000 jobs, split the query with facets.

The facets array lists filters with value ids and counts: job category, location, location country, worker type, time type. Add a value to appliedFacets as {"jobFamilyGroup": ["<id>"]}, using the facet's facetParameter as the key, and each slice is a fresh query with its own 2,000 rows. Choosing the facet matters:

• Prefer a facet whose counts add up to about the board total (job family, worker type). A facet where many jobs have no value is unreachable for those jobs.
• A multi-valued facet such as location counts a job once per location, so slices overlap; deduplicate by id.
• Prefer slices that all stay under 2,000, then the fewest slices. If a slice is still over the cap, split again with a second facet, but limit the depth to avoid loops.

Python

import requests

BASE = "https://nvidia.wd5.myworkdayjobs.com/wday/cxs/nvidia/NVIDIAExternalCareerSite"
CAP = 2000

def query(applied, offset):
    body = {"appliedFacets": applied, "limit": 20, "offset": offset, "searchText": ""}
    return requests.post(BASE + "/jobs", json=body, timeout=60).json()

def crawl(applied, seen):
    first = query(applied, 0)
    if first["total"] >= CAP:                       # over the cap: split by a facet
        facet = next(f for f in first["facets"] if f.get("facetParameter") and f.get("values"))
        for v in facet["values"]:
            if v.get("id"):
                crawl({**applied, facet["facetParameter"]: [v["id"]]}, seen)
        return
    for offset in range(0, first["total"], 20):
        page = first if offset == 0 else query(applied, offset)
        for p in page["jobPostings"]:
            seen[p["externalPath"]] = p
seen: dict = {}
crawl({}, seen)
print(len(seen), "jobs")

The detail call

GET https://{tenant}.{wdN}.myworkdayjobs.com/wday/cxs/{tenant}/{site}{externalPath} returns jobPostingInfo: the description HTML, an exact startDate, location and additionalLocations, jobReqId, timeType, remoteType, the country code and the public externalUrl. It is one request per job, which is why Workday costs more than any other system: a first read of a board with 1,400 jobs is about 70 list calls plus 1,400 detail calls. Fetch details for new ids only. Pay ranges are written into the description text, not a field.

Pace and politeness

Workday returns 429 when you hit many tenants quickly. In our crawl the budget looked like a per-IP limit across tenants, not per host: several tenants in parallel at a few requests per second ran into 429s, and two crawlers on one IP add up. Keep a few requests per second in total, back off with a growing pause on 429, run one crawler per IP and use the delta approach (details for new ids only). The employer name inside a posting (hiringOrganization) is a legal entity or site ("IN03 NVIDIA Graphics, Hyderabad"), so take the brand from the tenant, not from the posting. Boards also change address when a company migrates data centres, so keep the full host and site in your key, not the tenant alone.

The fast route: counts and postings per company, no tenant needed

Arcadium Data crawls Workday boards in deltas, handles the facet split, keeps tenant, data centre and site in the board key, and serves each company by domain. The company call is open without a key:

Calls on this page work without a key from one IP (10 requests a minute, 100 a day, short answers, point lookups). A free key lifts the limits: sign up. Routes, parameters and fields: OpenAPI.

curl

curl -s "https://api.arcadiumdata.io/v1/companies/nvidia.com"

Python

import requests

data = requests.get("https://api.arcadiumdata.io/v1/companies/nvidia.com", timeout=30).json()
h = data["hiring"]
print(h["ats"], h["boardUrl"], h["openJobs"], h["jobsPostedLast30Days"])
print(data["signals"]["openings7d"], data["signals"]["closures7d"])

TypeScript

const res = await fetch('https://api.arcadiumdata.io/v1/companies/nvidia.com');
const data = (await res.json()) as { hiring: { ats: string; boardUrl: string; openJobs: number; jobsPostedLast30Days: number }; signals: { openings7d: number; closures7d: number } };
console.log(data.hiring.ats, data.hiring.openJobs, data.hiring.jobsPostedLast30Days);
console.log(data.signals.openings7d, data.signals.closures7d);

Response (abridged)

{
  "domain": "nvidia.com", "name": "NVIDIA",
  "hiring": {
    "ats": "workday",
    "boardUrl": "https://nvidia.wd5.myworkdayjobs.com/NVIDIAExternalCareerSite",
    "openJobs": 2742, "jobsPostedLast30Days": 1062, "remoteShare": 0.184,
    "topCountries": [{ "name": "US", "share": 0.497 }, { "name": "IL", "share": 0.15 }],
    "matchConfidence": "high"
  },
  "signals": { "openings7d": 384, "closures7d": 18, "countries": 32 }
}

That one call replaces finding the three address parts, the facet split (this board is over the cap) and a detail call per job. With a free key, GET /v1/jobs with ats=workday lists the postings across companies (filters country, q, postedSince). The market view, with Workday's share and its switches in and out, is GET /v1/ats/market and GET /v1/ats/switches.

Comparison

Related

• Companies using an ATS
• Which ATS does a company use
• Posted salary data from job ads
• SmartRecruiters and Personio endpoints
• ATS vendors

Sources

• Workday
• Common Crawl URL index

Last updated 2026-10-25. This guide is ours and independent of the organisations it describes. Found a mistake? Tell us and we correct it.

Cite as: Arcadium Data, "Workday career site JSON: the jobs endpoint, facets, the 2,000-row cap and big boards", https://arcadiumdata.io/guides/workday-career-site-json-api/, data as of 2026-10-25.