Remote Jobs API: Match & Score

Aggregate current remote-job listings from four public APIs, normalize and deduplicate them, then rank each role against a candidate or hiring profile with explainable deterministic scoring.

Sources

• Arbeitnow
• Jobicy
• Himalayas
• Remote OK

No login, browser, paid data provider, or external LLM is required. Each result preserves source, source_url, apply_url, and attribution. Remote OK results link back to Remote OK as required by its public API terms.

Features

• live HTTP fetch with retries for rate limits and transient server/network failures
• partial-failure tolerance: one broken feed does not discard the others
• normalized title, company, description, location, employment type, salary, currency, skills, seniority, dates, and URLs
• cross-source deduplication by canonical URL and normalized title + company
• filters for keywords, region, country, employment type, salary floor, and posting age
• explainable fit score, fit band, reasons, and data-quality risk notes
• one dataset item per unique ranked job
• run diagnostics in the OUTPUT key-value-store record

Input example

{
  "platforms": ["arbeitnow", "jobicy", "himalayas", "remoteok"],
  "search_keywords": "python",
  "max_jobs_per_platform": 10,
  "max_total_jobs": 40,
  "profile": {
    "desired_roles": ["python engineer", "backend engineer"],
    "skills": ["python", "django", "postgresql"],
    "seniority": "senior",
    "regions": ["europe", "worldwide"],
    "timezone": "UTC+2",
    "salary_min": 80000
  },
  "posted_within_days": 30
}

search_keywords uses AND semantics after normalization: all words must appear somewhere in the title, company, description, or skills.

Output

Each dataset item includes:

• fit_score, fit_band, fit_reasons, risk_notes
• title, company, description, location, remote_region
• employment_type, salary_min, salary_max, currency
• skills, seniority, published_at
• apply_url, source, source_url, attribution

Known salary values below salary_min are discarded. Listings without salary data are retained and marked in risk_notes rather than silently removed.

Pay-per-event

• run-start: once per valid run.
• job-scored: once per unique emitted job.
• Discarded, duplicate, or failed-source rows are not charged.

Local verification

uv run --with-requirements requirements.txt --with pytest python -m pytest tests -q
uv run --with-requirements requirements.txt python -m src.main \
  --input samples/input.json \
  --output /tmp/remote-jobs-output.json
apify validate-schema