Poslovi Infostud Jobs | Normalized

Search stored Poslovi Infostud jobs with complete original descriptions, optional owner-funded AI enrichment, optional English display text, and advanced filters. This independent Actor is not affiliated with Poslovi Infostud.

How it works

A private collector reads public source data and stores reliably parsed facts and complete descriptions. This public Actor queries that shared inventory. It never searches or requests the Poslovi Infostud website during your run.

A job is available only when the collector has verified its full description. Snippets, truncated bodies, and unverified historical records are withheld. If the inventory is unavailable or stale, the run fails instead of falling back to website scraping. Coverage can therefore be smaller than the current website listing.

When requested, AI enrichment fills supported unknown fields and records provenance in llm. It cannot overwrite known source facts. Jobs whose requested enrichment fails are withheld. Optional translation converts selected human-readable fields into English, while raw.description and raw.descriptionHtml preserve the original source content. Without enrichment, llm.status is not_requested. Visible description markup is preserved in raw.descriptionHtml; hidden tracking text and reference markers are removed by the collector. Image-only adverts without a complete readable description are withheld.

First run

{
  "schemaVersion": "nomad-agent-inventory-search-v1",
  "maxItems": 5,
  "postedWithin": "any",
  "keyword": "",
  "aiEnrichment": {"enabled": false, "accuracy": "silver"},
  "translateToEnglish": false,
  "dedupe": {"enabled": false, "key": ""}
}

maxItems is an upper bound of 200; zero also means 200. Filtering, repeat suppression, and failed enrichment can produce fewer results. A keyword is a case-insensitive phrase matched against the stored title, employer, and complete original description. It is never sent to the source website.

postedWithin accepts any or a positive duration such as 7d. When Poslovi Infostud has no reliable original posting date, the stored date is the job's first admission to the inventory. Later refreshes do not make the posting appear new.

Advanced filters

workArrangements accepts remote, hybrid, and onsite. For other normalized fields, use the shared nomad-agent-job-filter-v1 expression contract:

{
  "schemaVersion": "nomad-agent-job-filter-v1",
  "expression": {
    "field": "data.compensation.minimum",
    "operator": "gte",
    "value": 50000
  }
}

Place that object under the input's filters field. With AI disabled, filters evaluate source facts only; unknown values do not satisfy ordinary comparisons. With AI enabled, known source mismatches are removed before AI calls and unknown enrichable values are checked again after enrichment. Filters operate on normalized values before English display translation. Compare compensation only with the appropriate currency and period filters.

Output and repeat delivery

Every dataset item follows nomad-agent-job-v1, with six roots: schemaVersion, identity, data, llm, raw, and custom. Missing facts remain null; a known empty collection is []. Original identity and complete description evidence remain stable across enrichment and translation. The rich format is intended for v3 scoring and advanced filtering.

The default key-value store contains RUN-SUMMARY, including inventory freshness, candidates, withheld enrichment failures, delivery counts, and terminal status. OWNER-USAGE separates actual provider costs from unpriced token counts when enrichment runs. Diagnostic receipts are not inserted into the jobs dataset.

Repeat suppression is enabled by default. It uses a persistent ledger scoped to the Apify user, Actor, and search. Set dedupe.enabled to false for a stateless run. A nonempty dedupe.key deliberately shares delivery history within that user's scope. Ambiguous delivery or ledger failures fail the run rather than claiming successful delivery.

Owner-funded processing

AI enrichment and translation are independent opt-ins, both off by default. Enable aiEnrichment.enabled for extraction and translateToEnglish for English display text. You do not supply API keys. Only when enabled, the owner sends description text to OpenRouter for extraction or selected display text to DeepL for translation. The configured routes deny provider data collection but do not guarantee zero retention or EU-only processing. Sanitized extracted facts and translations may be cached by the owner. V3 callers can explicitly request the processing they need.

The existing Apify pricing configuration is preserved by this migration. This Actor is unmonetized; platform compute or storage charges can still apply. No new per-result, enrichment, or translation price is introduced.

Use nomad-agent-inventory-search-v1 for this database-only release. The previous source-fetch input required infostud.startUrls; those source URLs are not accepted by the inventory search and are never silently discarded. Migrate the request to keyword or supported normalized filters. Original descriptions are always included.

The top-level accuracy field remains a compatibility override for aiEnrichment.accuracy; it never enables enrichment by itself.

Relative posting-age windows are limited to 36500 days. Use any for an unrestricted search. AI enrichment and translation remain independent opt-ins, both off by default.

Source-specific facts use the public Infostud schema.

Job Atlas

Explore the Job Atlas job-data and matching Actors. Use latest and retain the immutable build ID returned by each run.

• Linkedin
• Euraxess
• YC
• Scorer

Product guide | Source and client examples