AI Job Search & Fit Scorer — 10 Sources + V3 Matching
Pricing
$20.00 / 1,000 retained job fit results
All-in-one developer job search and fit scoring: search 10 public sources or score your own jobs against a résumé or profile. Get a gate-aware shortlist with fit evidence, skill gaps, hard-requirement checks, and application links.
Choose Search + score to find current jobs, or Score supplied jobs to evaluate normalized jobs you already have.
Used only in Search + score. The fresh form searches three public sources for software engineer roles and fetches at most three jobs per source. Edit sources, terms, and filters for your market. Collectors run inside this Actor—there are no paid child Actor runs or stacked child fees.
{ "sources": [ "linkedin", "remote_boards", "justjoinit" ], "keywords": [ "software engineer" ], "postedWithinDays": 14, "maxItemsPerSource": 10, "cacheTtlSeconds": 1800, "concurrency": 3, "sourceTimeoutSecs": 120}JSON array of complete nomad-agent-job-v1 records. Every item needs top-level schemaVersion, identity, data, custom, llm, and raw; identity needs source plus externalId or url, and raw.description contains the posting text. Copy the working shape from README → Score jobs you already have or examples/inline-input.json. Use this, sourceDatasetId, or sourceActorRunId—never more than one. Hard limits: 200 records and 20,000 plain-text description characters per job; oversized evidence is rejected, never truncated.
Paste the immutable Apify dataset ID—not the dataset name or URL. Each dataset item must be one complete nomad-agent-job-v1 record. The dataset is read only and never modified. Use only one supplied-job source.
Paste the exact upstream run ID—not the Actor ID, build ID, or run URL. The run must be terminal SUCCEEDED with exit code 0; this Actor reads that run's default dataset as nomad-agent-job-v1 records. Use only one supplied-job source.
Optional guard for sourceActorRunId. Enter the upstream run's exact immutable build ID or build number; the run fails instead of reading jobs when it does not match.
Maximum unique jobs to evaluate (1–200). In search mode this caps the merged, deduplicated result set; in score-jobs mode it caps the supplied source.
JSON object with at least one of primaryRole, targetTerms, skills, or freeText. Optional seniority, experience, language, location, and contract constraints must be evidence-backed. The displayed example is illustrative and is not prefilled. Use exactly one of this, Résumé, or Résumé text.
{ "primaryRole": "Software Engineer", "targetTerms": [ "Software Engineer", "Backend Engineer" ], "skills": [ "Python", "TypeScript", "SQL", "Docker" ], "seniorityLevels": [ "mid" ], "yearsExperience": 4, "workableLanguages": [ { "language": "English", "level": "professional", "evidence": "Fake demo profile: professional working proficiency" } ], "workableLanguagesComplete": false, "acceptedWorkArrangements": [ "remote" ], "remoteLocations": [ "Worldwide" ], "hybridLocations": [], "onsiteLocations": [], "workArrangementPreferencesComplete": false, "acceptedContractTypesComplete": false, "freeText": "Fake demo candidate for testing the Actor. Replace every value with your own evidence before using the results."}Text-based PDF or UTF-8 text, up to 5 MB, 50 PDF pages, and 8,000 readable characters. It is read in memory and sent once for structured profile extraction; it is not written as a separate dataset artifact or repeated in each fit prompt.
Plain-text alternative for API clients, up to 8,000 readable characters. Use exactly one of candidateProfile, resume, or resumeText.
Up to 4,000 characters of explicit role, location, work-mode, language, contract, or other constraints. Unknowns are never guessed.
Shortlist (default) returns and charges only successfully scored rows whose deliveryScore meets minDeliveryScore. Audit returns the complete evaluation dataset, including static_drop, static_hold, forward_cap_hold, and ai_failed rows; as before, every Audit row except ai_failed is charged.
Inclusive 0–5 output threshold used only in Shortlist mode. Default 2. deliveryScore includes hard-gate reductions, so it is safer for delivery than raw fitScore alone. Audit mode records this value but does not filter by it.
V3 static candidates below this 0–100 rank are held before AI (default 30). Hard contradictions are always dropped.
Cap AI verdict calls after static selection (0–200). Remaining eligible jobs receive forward_cap_hold rather than a fabricated score.
Also evaluate v3 recoverable holds. Disabled by default so uncertain static evidence remains visibly held.