Dice Tech Jobs Scraper (USA)

Tech job listings from Dice.com, one of the largest US tech-specific job boards (part of DHI Group). Search by keyword and optional location; every job comes back with title, company, location, salary, employment and workplace type, posting date and a description snippet — from a single search call, with an optional pass for the full job description.

No API key, no login, HTTP only.

Example input

{
  "keywords": ["Python Developer", "DevOps Engineer"],
  "location": "Austin, TX",
  "includeJobDetails": false,
  "maxItems": 100
}

What you get

Records share one dataset, told apart by recordType:

• JOB — one per listing. Carries the envelope plus jobId, jobUrl, keywordQueried, locationQueried, searchPage/searchRank, and the raw search entry under listing — passed through verbatim, so upstream's own field names are preserved:

  field	example
  title	Python Developer
  companyName	The Avian Consulting LLC
  jobLocation	{ "city": "Hartford", "state": "Connecticut", "country": "USA" }
  salary	$60 - $70 (as Dice prints it — hourly or annual, sometimes a range)
  employmentType	Full-time
  workplaceTypes	["On-Site"] · also isRemote, workFromHomeAvailability
  postedDate / modifiedDate	ISO 8601 UTC
  summary	a description snippet
  detailsPageUrl · guid	the job's own Dice URL and id
  companyLogoUrl · easyApply · employerType

  With includeJobDetails on, jobDetail is filled from the job page's JobPosting structured data, which carries the full HTML-formatted description (the search snippet is only a teaser).

• SEARCH_SUMMARY — one per keyword: totalResults reported by Dice, returnedCount, pagesFetched, and locationApplied (see the location note).

• ERROR — one per keyword that failed, so every keyword maps to at least one row.

Every row carries _input, _source, _scrapedAt, recordType.

Input

Field	Meaning
keywords	Required. One full-text search per entry (Python Developer, DevOps Engineer).
location	Optional, applies to every keyword. Format as Dice shows it: Austin, TX, New York, NY.
includeJobDetails	Fetch each job's page for the full description. Off by default — it is one extra request per job.
maxItems	Stop a search after this many jobs (default 100; 0 = unlimited, still bounded by maxPages).
maxPages	Pagination-depth cap (default 20; 20 jobs per page).
maxConcurrency	Requests in flight at once (default 5).
minRequestInterval	Seconds between request starts (default 1). Dice showed no rate sensitivity in recon; kept small as a courtesy.
proxyConfiguration	Apify US residential by default.

Known limits — worth reading

• US-focused. Dice is a US tech board; a non-US location mostly returns little or nothing. That is the source, not a bug.
• An unrecognised location is silently ignored by Dice — the search runs nationwide instead of erroring. locationApplied on the summary row tells you whether the location actually bound, so you can tell "nationwide" apart from "filtered".
• The summary field is a snippet, not the full posting. Turn on includeJobDetails for the complete description (jobDetail).
• salary is exactly what Dice prints — sometimes hourly, sometimes annual, sometimes a range or blank. It is passed through as text, not normalised.
• Roughly 900+ results per broad keyword. Results thin out past the first pages; the actor stops when a page stops adding new jobs rather than paying for Dice re-serving the last page (which it does instead of erroring at the end).

Method

Dice's robots.txt disallows the query-string search form (/jobs?q=…) that its own search box uses, but allows the path-based search form its sitemap publishes for SEO (/jobs/q-{keyword}-jobs). This actor uses only the allowed path form. Search results are read from the page's Next.js RSC stream; the full description on the detail pass comes from the JobPosting JSON-LD block. No WAF or bot-mitigation gate was seen in recon — six TLS profiles answered 200 cold.