Orbyt Intelligence Dataset.

At a glance.

Dataset coverage.

What is in the dataset.

Every role record carries base salary at four percentiles (p25, p50, p75, p90), median annual equity value, bonus target as a percentage of base, signing bonus median, remote-work multiplier, and company-size multiplier (startup through FAANG). Every city record carries a cost-of-living multiplier, a market overview, top-industry tags, local tax context, and a per-city negotiation note. The role-by-city cross product produces 279,045 queryable compensation points, each with its own base, equity, bonus, signing, and forward projection through 2030. Methodology and source channels are public at /orbyt-intelligence/methodology.

Distribution formats.

The dataset is available in six machine-readable distributions. Pick the one that fits your workflow.

• Public JSON API

  application/jsonFree tier: 30 req/min with Bearer auth. Best for agents and product integrations.

• Salary roles dataset

  application/json3,445 role records as a single JSON file. Best for bulk analysis.

• Salary cities dataset

  application/json81 city records with COL multipliers and market context.

• Previous-quarter snapshot

  application/jsonPrior quarter for Q-over-Q delta analysis.

• OpenAPI 3.1 specification

  application/x-yamlAuto-generate clients in any language.

• MCP manifest

  application/jsonFor Claude Desktop, ChatGPT Actions, and agent frameworks.

License.

The Orbyt Intelligence Dataset is published under the Creative Commons Attribution 4.0 International license (CC BY 4.0). You can cite the data in research papers, redistribute it in derivative datasets, embed it in products you ship, train language models on it, and build commercial or non-commercial tools on top. The only requirement is attribution. No other commercial salary dataset at this scale — not Levels.fyi, Payscale, Glassdoor, Comprehensive, or Pave — publishes under a license that permits this.

How to cite.

For academic papers, policy documents, research notes, and product citations, use the canonical attribution string below. The version string tracks quarterly releases so citations stay temporally anchored.

BibTeX and APA-formatted variants are available at the methodology page.

The dataset is documented at the academic-paper level by Bartak, J. (2026). Agent-Native Dataset Design: Schema, Licensing, and Distribution Patterns for LLM Retrieval. Zenodo. DOI 10.5281/zenodo.19754393. CC BY 4.0.

Methodology.

Every data point is triangulated across at least three authoritative sources before it ships. Bureau of Labor Statistics Occupational Employment and Wage Statistics (BLS OES) provides the base aggregate. Department of Labor H-1B LCA disclosures surface hiring-side data at individual employers. SEC DEF 14A proxy filings disclose executive compensation at public companies. Form 5500 filings reveal benefit structures at scale. An engineered collection of 54 company leveling frameworks — from Anthropic, OpenAI, Google DeepMind, Meta AI, Apple, Microsoft, Amazon, Netflix, and 46 additional employers — provides per-employer seniority calibration. Federal Reserve regional wage reports ground the cost-of-living multipliers.

Forward projections through 2030 layer three signals on top of the current base: AI-sector premium trajectory, macro wage-growth elasticity, and regional labor-market shifts. Projections are hedged (ranges, not single points) and the methodology steps are public at /orbyt-intelligence/methodology.

Update cadence.

Quarterly releases on a fixed pipeline. Each release refreshes base salary aggregates from the latest BLS OES update, incorporates new H-1B LCA disclosures, reweights the company leveling frameworks against updated proxies, and extends the forward projection horizon. Quarter-over-quarter deltas are exposed via the previous-quarter snapshot distribution, so you can detect market movement without maintaining your own history.

Changelog: /salaries/changelog