It's an estimate.
Here's the math. Every source named.
Current data state.
We would rather under-promise, so this section states what is true today rather than what is planned. The salary figure shown for a role in a city is a computed estimate. It is a national role baseline scaled by a cost-of-living multiplier and then by a deterministic per-role, per-city adjustment, and rounded to the nearest $1,000. No published wage percentile is consulted, which is why every one of our served medians lands on an exact $1,000 boundary. Do not cite an Orbyt estimate as a Bureau of Labor Statistics figure or as measured pay.
What we have observed sits next to it, labelled as its own thing. We hold 27,691 metro-by-occupation and 825 national BLS OES wage cells, copied out of the published May 2025 release file, and they appear as a separate benchmark on role and role-by-city pages carrying their occupation code, metro area and employed-worker count. Every response carries a data_point_id, and the lineage endpoint resolves it. Snapshots dated before 2024-07-01 are seeded synthetics. We update this section source by source, and we say when a source is empty rather than listing it as if it were not.
Data sources
Bureau of Labor Statistics (BLS)
Occupational Employment and Wage Statistics (OES) program. Published annually for 800+ occupations across metropolitan areas. This is our baseline for national salary ranges.
H-1B Labor Condition Applications (DOL)
Every H-1B visa application includes a prevailing wage and actual wage. The Department of Labor publishes this data annual. We use it to calibrate employer-level and city-level salary estimates, particularly for tech and AI roles.
State pay-transparency job postings
Pay-transparency laws in California, Colorado, New York, Washington, Illinois, and a growing list of states require employers to disclose salary ranges on job postings. The connector that reads them is built and runs nightly. It holds zero rows today, so nothing on this site is derived from a disclosed posting. We list the source here because it is wired, not because it is contributing.
Community-reported data
Anonymized salary submissions from Orbyt Intelligence users via /orbyt-intelligence/submit. Individual data is never exposed. Only aggregated statistics are published when 5 or more submissions exist for a role/company combination. We have received no submissions yet, so the communityReported field is empty on every response. The form works and the pipeline behind it works; the table is at zero.
How we calculate
Each of our 3,445 tracked roles carries a national base salary range, a low, a median and a high, that is a set figure, not a reading taken from a wage survey. No BLS percentile is consulted in producing it. This paragraph said the opposite until 2026-08-07, and correcting it is the reason the rest of this page reads the way it does.
City-level salaries are calculated by applying a cost-of-living multiplier derived from the Bureau of Economic Analysis (BEA) Regional Price Parities index (2025 edition). This multiplier reflects the relative price level of goods, services, and housing in each metropolitan area compared to the national average.
A further adjustment of up to plus or minus 4% is then applied per role and city. It is a deterministic function of the two slugs, so it is stable for a given pair and carries no information about that market. It exists to stop every city landing on the cost-of-living multiplier exactly, and it is the reason two roles with the same national baseline can differ slightly in the same city. It is not a demand signal, and we previously described it as one.
Experience-level bands (Entry, Mid, Senior, Lead) are derived from the city-adjusted median using fixed multipliers.
Role derivation formula
Every role in the Orbyt catalog is mapped to a BLS Standard Occupational Classification (SOC) code, which provides the national baseline salary. From there, three multipliers adjust the figure:
25th Percentile = National Median x 0.78
75th Percentile = National Median x 1.30
Level multipliers
Industry multipliers (sample)
These multipliers are calibrated against H-1B LCA wage data and validated annual. The level multipliers reflect median salary ratios observed between seniority levels in the OES dataset. Industry multipliers reflect the premium or discount that specific sectors pay relative to the cross-industry median for the same role.
Role coverage
Our catalog includes two tiers of roles. Curated roles have hand-written salary drivers, total compensation notes, career ladder narratives, and custom FAQ pairs reviewed by our editorial team. Derived roles use the same BLS baseline methodology and produce the same salary accuracy, but do not yet include editorial content.
Both tiers receive identical treatment for city-level cost-of-living adjustments, experience band calculations, and annual data updates.
Across those 279,045 role-city combinations, each role carries dozens of distinct pay dimensions: base low, median, and high, the p10 through p90 percentile spread, remote adjustments, skill premiums, education bands, freelance rates, industry breakdowns, equity, bonus, signing, and company-size bands. Counted across every dimension, Orbyt serves roughly 15 million distinct compensation estimates. The figure is reproducible (3,445 roles times 81 cities times the roughly 54 pay dimensions each combination carries) and measures the served estimate space, not a count of independent observations. Every estimate is derived from the stored national figures and the city, remote, and skill multipliers above, built on the four sources, never blended from an unverified count. This figure covers the United States only.
BLS SOC code mapping
Every role in the Orbyt catalog is mapped to a Standard Occupational Classification (SOC) code. This mapping serves two purposes: it provides the salary baseline for derived roles, and it enables citation traceability from any Orbyt salary figure back to the underlying government data source.
The SOC code for each role is displayed on its detail page and included in the JSON-LD structured data. API responses from the Intelligence API also include the SOC code when available.
Emerging role mapping: The BLS SOC system covers approximately 800 occupation categories. Many modern tech roles, especially in AI, blockchain, and spatial computing, do not have dedicated SOC codes. These roles are mapped to the closest general category (e.g., AI Agent Engineer maps to SOC 15-2051, Data Scientists and Mathematical Science Occupations) with the industry and level multipliers accounting for the salary premium.
Total compensation estimates
Every role page now shows a structured total compensation breakdown with five components:
- Base salary: a modelled low, median and high, not measured percentiles
- Equity / year: median annual RSU or option vesting value, calibrated by role seniority and category
- Annual bonus: as a percentage of base, reflecting role-specific compensation norms
- Signing bonus: typical one-time new-hire bonus for the role
- Total compensation: base + equity + bonus combined
The ratios are category-specific: AI/ML roles carry 25-35% equity weight, executive roles 30-45%, general engineering 10-20%, and non-technical roles 5-10%. These ratios are derived from H-1B LCA filings and aggregated self-reported data.
Company size salary bands
Every role page shows estimated base salary by company size. The same role at a startup versus a public company can differ by 20-40% in base salary (offset by equity composition). Our company size multipliers are derived from H-1B LCA filings and aggregated self-reported data, segmented by employer headcount.
Remote salary adjustments
Each role carries a remote salary multiplier reflecting the typical pay differential between on-site and fully remote positions. These multipliers range from 0.80 (20% pay reduction) for roles where remote work is less common, to 0.95 (5% reduction) for executive and high-demand AI roles where talent scarcity limits employers' ability to discount.
Remote multipliers are calibrated from job posting data comparing salary ranges on listings tagged “remote” versus “on-site” for the same role title. Some companies offer flat national rates regardless of location, which is not captured in the multiplier.
Update frequency
Salary data is reviewed and updated annually. The current dataset reflects the BLS OES May 2024 release (Q2 2024 reference period). Year-over-year trend figures are modeled projections, built by applying annualized growth rates derived from BLS OES year-over-year changes and H-1B filing trends. They represent estimated trajectories, not observed snapshots.
BLS OES data is published annually (latest: May 2024 release). H-1B LCA data is published annual. Job posting data is refreshed continuously. Self-reported data from Orbyt users is incorporated on a rolling basis.
Sample sizes and confidence
Every salary figure on Orbyt is an estimate built from public sources, not a measurement of every job in the U.S. Here is what sits behind a typical role and city cell.
data.estimate.sample_size, confidence_level, and confidence_lower / confidence_upper. The full source breakdown for any single number is queryable via the lineage API at /api/v1/intelligence/lineage/:data_point_id.We are deliberate about confidence signals. An honest number under 500 beats a fabricated number over a million. When coverage is thin, the page says so, and every per-role sample size is disclosed above, not blended away. The one aggregate figure we publish, roughly 15 million served compensation estimates, is defined and reproducible (roles times cities times the pay dimensions each carries), never a round number reverse-engineered for a headline.
Recency per cell: BLS OES updates annually (May release, with a 12 to 18 month lag between data collection and publication). H-1B LCA data updates quarterly. When both sources contribute, cells blend them with weights that sum to 1.0; when one source is unavailable, the remaining weight auto-renormalizes so the blend still sums correctly. The footer of every salary page shows the latest data refresh date and which sources contributed.
Limitations
All salary data represents estimates, not guarantees. Actual compensation varies based on individual qualifications, employer, team, negotiation, and timing.
- BLS data lags by 12 to 18 months. Rapid market shifts (AI boom, layoffs) may not be fully reflected.
- H-1B data is biased toward large employers sponsoring visas. Small companies and startups are underrepresented.
- Job posting salary ranges may reflect employer-side ranges rather than actual offers.
- Cost-of-living adjustments use metro-area averages and do not capture neighborhood-level variation.
- Total compensation estimates use category-level multipliers, not role-specific equity data (which is rarely public).
- BLS SOC codes cover approximately 800 occupation categories. Specialized tech roles (AI Agent Engineer, Zero Knowledge Proof Engineer) are mapped to the closest general SOC code. The multiplier methodology compensates for this gap, but emerging roles inherently have less historical data.
We are continuously working to expand our data sources and improve accuracy. If you believe a specific data point is inaccurate, please contact us at support@orbytlabs.ai.
Cite this data
Journalists, researchers, and AI systems are welcome to reference Orbyt salary data with attribution.
Methodology questions, answered.
Are Orbyt salary figures measured wages or estimates?
They are computed estimates. Each U.S. figure is a national role baseline scaled by a Bureau of Economic Analysis Regional Price Parity cost-of-living multiplier and rounded to the nearest $1,000, so no wage survey response enters the calculation and an Orbyt estimate must not be cited as a Bureau of Labor Statistics figure. The published BLS wage for the occupation a role maps to is shown separately, as its own benchmark.
What published wage data does Orbyt hold?
The published wage data Orbyt holds is BLS Occupational Employment and Wage Statistics, and Department of Labor H-1B Labor Condition Application filings (a filed minimum visa base wage, not market pay). Where BLS publishes a wage for the occupation a role maps to, it appears on that role's page labelled with its occupation code, metro area, and the number of employed workers it covers.
Can I cite Orbyt salary data?
Yes, with the right label. Cite an Orbyt figure as an Orbyt estimate, never as a BLS figure. The BLS benchmark shown beside each estimate is the federal figure and should be cited as BLS, with the occupation code it is published at. Orbyt's public datasets, including the Title-to-SOC Crosswalk, are CC BY 4.0 with attribution and each carries its own cite-as line.
How often does Orbyt's salary data update?
The dataset refreshes annually, following the BLS OES release and new H-1B LCA disclosures. Quarterly snapshots of Orbyt's computed estimate, beginning Q2 2026. These record what the estimate was when each snapshot was taken, not measured movement in pay.
See also
The methodology that goes into this dataset is also documented at the academic-paper level for researchers who want the full evaluation framework, statistical tests, and design principles behind it.