Methodology
How we compute our data
Every figure on Landed is computed from live AI-native job postings — not surveys. We normalise roles to canonical titles, use posted-band midpoints, annualise equity over a four-year vest, report percentiles, and only publish pages backed by enough real data to be meaningful.
Where the data comes from
Every salary, open-role count, and company signal on Landed is computed from live job postings at AI-native companies. We do not use self-reported salary surveys, which lag the market and skew toward whoever chooses to respond. Postings reflect what companies are actually offering right now, and our figures move as the market does.
How we normalise roles
Job titles are messy — "AI Engineer," "ML Engineer," "Applied AI Engineer," and a dozen variants overlap. We map postings to a set of canonical roles so that comparisons are apples-to-apples, while preserving the original title on each individual posting. Seniority is inferred from the posting (title, scope language, and posted band) into Mid / Senior / Staff / Principal tiers.
How we handle compensation
- Where a posting lists a salary band, we use the midpoint.
- Where it lists equity, we annualise it over a standard four-year vest and add it to base to estimate total compensation.
- We report percentiles (25th / median / 90th) rather than a single number, because the spread matters as much as the midpoint.
- Figures are calibrated ranges, not offers — actual compensation varies with location, stage, performance, and the share price at each vest.
The data-thickness rule
We only publish a salary, role, or location page when the underlying postings are numerous enough to be meaningful. If a role-by-city or company-by-role combination has too few postings to report responsibly, we don't ship a thin page for it — we either omit it or mark it clearly. This keeps every page backed by real data rather than padded to fill a template.
How often it updates
Job and company data refreshes continuously; aggregate pages (salaries, roadmaps, the report) are recomputed on a regular cadence. Each page shows a "last updated" date so you always know how fresh the numbers are.
Limitations
No dataset is perfect. Postings can be reposted or left up after a role is filled; equity is inherently uncertain; and some companies post bands wider than what a typical offer lands at. We surface ranges and sample sizes precisely so you can weigh the numbers rather than treat any single figure as gospel. If you spot something off, we want to know.
See the data in action
Explore live AI-native salaries, jobs, and company hubs — all computed the way described above.