Dice Jobs Scraper: US tech job search results from dice.com

Give it a keyword (and optionally a location) and get the jobs Dice lists for that search as clean JSON: title, employer, location, employment type, on-site/remote/hybrid, salary when the employer published one, posting date, a short excerpt and the link to the posting. Each run also tells you what changed since the last run: new jobs, jobs still listed, and jobs that dropped off.

• US tech jobs: Dice is the US technology job board. Enter python, data engineer in Austin, TX, devops in Texas, or paste a Dice results URL.
• Plain requests to public pages, nothing else: one request per results page, the same page a visitor opens. No login, no proxies, no CAPTCHA or anti-bot workarounds, and the site's robots.txt is checked and obeyed during the run.
• Structured data, not scraped HTML: the results page embeds its job list as data (the Next.js page payload); the Actor reads that, so fields are consistent and do not break when the page design changes.
• Only robots-allowed URLs. Dice's robots.txt forbids crawlers from using the query-style search (/jobs?q=...) but allows the path-style pages (/jobs/q-python-jobs, /jobs/q-python-l-Austin%2C_TX-jobs) and their ?page=N continuation. The Actor builds only those. By default each search returns its first page, 20 jobs; raise Max jobs per search to read more pages (up to the Actor's limit of 25 pages).
• Change tracking built in: every record has changeStatus (new or existing) and firstSeenAt. Turn on Only new jobs and schedule the Actor daily to get a feed of what appeared.
• Reliable and polite: timeouts, retries with backoff on server errors, at most two requests at a time and one per second, a descriptive User-Agent. A bad search is reported in the run summary and the rest still runs.

Use cases

• Daily job alerts: "python in New York, NY", "react in Remote", "sre in Seattle, WA": run every morning with Only new jobs on and pipe the result into Slack, e-mail or a sheet.
• Recruiters and staffing agencies: watch which employers (and which competing recruiters) post for your specialities, with employment type, Easy Apply and visa-sponsorship flags.
• Labour-market and salary research: sample postings per keyword and location, including published pay and workplace type; searchTotalResults gives Dice's total count per search for trend lines.
• Sales intelligence: hiring is a buying signal; find companies opening "Snowflake", "Kubernetes" or "Salesforce" roles in your region this week.

How to use it

1. Add searches to Searches, one per line, in any of these forms:
   • python (keyword, all of the US)
   • data engineer in Austin, TX (keyword and city; devops in Texas for a state; python in Remote works too)
   • https://www.dice.com/jobs/q-python-l-New_York%2C_NY-jobs (a results URL without ?...)
2. Set Max jobs per search if you want more than the first 20 (20 per page, one request per page).
3. For daily monitoring, schedule the Actor with Only new jobs turned on.
4. Run it. Jobs land in the dataset; a run summary lands in the OUTPUT record.

Keywords are matched the way Dice's own URLs work: words are searched as a phrase, and a hyphen counts as a space (front-end is searched as front end, because Dice reads q-front-end-jobs as front). Results come in Dice's default (relevance) order. A keyword Dice does not know returns no jobs, which is reported as a successful search with zero results.

Input example

{
    "searches": ["python in New York, NY", "data engineer in Austin, TX", "https://www.dice.com/jobs/q-devops-jobs"],
    "onlyNew": false,
    "maxItemsPerSource": 40
}

Field	Default	What it does
searches	1 sample search	Keyword, keyword in location, or a Dice results URL. One request per page read.
onlyNew	false	Outputs only jobs that earlier runs with the same searches have not seen.
maxItemsPerSource	20 (one page)	Reads at most N jobs per search, most relevant first; 20 per page, at most 500.

Output

One record per job. Fields are stable and documented in .actor/dataset_schema.json.

{
    "id": "dice:python/new york, ny:f997b73b-95a2-4b8d-b669-1396fdac00a2",
    "source": "dice",
    "sourceKey": "python/new york, ny",
    "title": "Senior Python Backend Developer (Microservices, FastAPI & GenAI)",
    "url": "https://www.dice.com/job-detail/f997b73b-95a2-4b8d-b669-1396fdac00a2",
    "company": "Rose International",
    "employmentType": "Full-time",
    "employerType": "Recruiter",
    "location": {
        "displayName": "Rutherford, New Jersey, USA",
        "city": "Rutherford",
        "state": "New Jersey",
        "stateCode": "NJ",
        "country": "USA"
    },
    "isRemote": false,
    "workplaceTypes": ["On-Site", "Hybrid"],
    "salary": "USD 80.00 per hour",
    "postedAt": "2026-10-08T00:01:03.000Z",
    "modifiedAt": "2026-10-08T00:01:03.000Z",
    "easyApply": true,
    "willingToSponsor": null,
    "descriptionExcerpt": "Date Posted: 10/07/2026 Hiring Organization: Rose International Position Number: 508796 Industry: Financial Services Job Title: Senior Python Backend Developer (Microservices, FastAPI & GenAI) Job Location: Rutherford, NJ, USA, 07070 Work Model: Hybrid Work Model Details: 3 days onsite and 2 days r…",
    "searchUrl": "https://www.dice.com/jobs/q-python-l-new_york%2C_ny-jobs",
    "searchPage": 1,
    "searchTotalResults": 5192,
    "firstSeenAt": "2026-10-08T06:50:56.000Z",
    "changeStatus": "new",
    "scrapedAt": "2026-10-08T06:50:56.817Z"
}

• employmentType: Dice's label(s), e.g. Full-time, Contract, Third Party, Part-time; several are comma-separated. employerType says who posted: Direct Hire (the employer), Recruiter, or Other.
• location: as Dice lists it; all fields are null when Dice lists none (typically remote jobs). isRemote and workplaceTypes (On-Site, Remote, Hybrid) come from the employer's own flags.
• salary: free text exactly as the employer wrote it (USD 120,000 - 150,000 per year, Depends on Experience); null when none.
• descriptionExcerpt: the first 300 characters of the posting as plain text. The full text is never output; follow url. Email addresses and common US telephone formats are redacted; summaries with contact, recruiter or hiring-manager labels are omitted.
• searchPage: which results page the job was on; searchTotalResults: Dice's total for the search, of which only the requested pages are read.

Run summary (OUTPUT record)

Totals (searches requested, succeeded and failed; jobs output, new, existing, removed since the last run), one entry per search with its status, a warning when a later page could not be read, and the list of errors. One bad search never fails the run. A search that Dice refuses (HTTP 403) is reported as such and never retried or worked around.

Change tracking: how it works

History is kept in a named key-value store, keyed by the set of searches. The first run outputs everything as new; later runs mark jobs new or existing, and the OUTPUT record lists IDs that dropped out of the pages read since the last run. With Only new jobs on, only new records are output. Jobs that disappear are remembered for 90 days, so a job that returns is not reported as new again. Because only the first N jobs of each search are read, "removed" means "no longer among the first N results", not necessarily "closed"; lowering Max jobs per search between runs therefore reports the jobs beyond the new limit as removed.

Pricing

Pay per event: one job event per record. Suggested price: $0.002 per job; free to try with the Apify free plan. The owner sets the actual price in the Store; it is never in the code.

What that costs, assuming every search returns a full page of 20 jobs and the price above:

• The default input (one search, one page) outputs 20 jobs: about 4 cents per run.
• Five searches with Max jobs per search at 100 (5 pages each) output 500 jobs per run, about $1.00; run daily, about $30 a month (30 runs).
• Five searches, one page each, with Only new jobs on: after the first run (100 jobs, about 20 cents) a daily run costs only the newly posted jobs, a few cents on a quiet day, up to 20 cents if every page turned over.

FAQ

• How many jobs can one search return? Up to 500 (25 pages of 20), which is the Actor's request limit. If Dice repeats or clamps a page, paging stops with a warning. The default is one page so a first run costs cents. Narrower searches (add a location, use a more specific keyword) cover more ground than deep paging.
• Can it filter by full-time, remote, posted date or salary? Not through Dice's filters, which are query parameters the Actor does not send (only ?page=N is used). Filter the output on employmentType, isRemote, workplaceTypes, postedAt and salary instead; the python in Remote form works for a remote-only search.
• Does it fetch the full job description? No. The results page already carries the opening of each posting; the Actor outputs its first 300 characters together with the link. Job detail pages are not requested.
• Does it log in or bypass bot protection? No. Plain requests with a descriptive User-Agent, robots.txt obeyed at run time. If Dice blocks a request, the run reports it and moves on.
• How fresh is the data? It is read live during the run, in Dice's default (relevance) order.
• Why front-end found "front" jobs? It does not: the Actor turns hyphens into spaces before building the URL, because Dice's path format uses the hyphen as its separator. Type front end or front-end, both search the phrase.

Legal

This Actor reads only public job postings that employers publish on Dice, sends plain GET requests to pages that Dice's robots.txt allows for every crawler, identifies itself with a descriptive User-Agent, obeys robots.txt at run time, requests nothing robots.txt disallows (no query-style searches, no /jobsearch/, no feeds, no APIs), and outputs a contact-screened excerpt of at most 300 characters plus the link. It stores no personal data beyond what the employer wrote into the posting. Dice's Terms and Conditions of Service restrict unapproved automated retrieval, republication and deep links; written permission remains an owner publishing decision; see LEGAL.md for the exact wording, dates and the open questions. "Dice" is a trademark of its owner; this Actor is not affiliated with or endorsed by Dice or DHI Group.