How Databricks Pays

Individual contributor ladder — Intern through L7

Management ladder — M2 through M8

Disclosed executive layer

Track divergence

Hierarchy qualifications and legacy structures

• The L3 to L7 and M2 to M8 labels come from compensation datasets and public requisitions, not from a Databricks policy document. The title and scope written into the specific requisition and offer letter are what actually control.
• Cross-functional equivalence is the weakest part of the map. The ladder is evidenced for software engineering; sales, solution architecture, field engineering and corporate functions can place equivalent scope on a different rung and pay it on a different architecture.
• Peer mapping is defensible by scope only, meaning entry, mid, senior, staff, senior staff or principal, manager, senior manager and director. No public source supports an exact level-number crosswalk to Snowflake, Confluent or Elastic.
• There is no public evidence of a predecessor band system or a post-acquisition harmonisation programme, so how acquired teams are slotted should be treated as unknown rather than assumed to follow the same ladder.
• University internships and new-graduate programmes are confirmed recruiting categories, but an intern band is not confirmed and the intern row here is an annualised hourly posting rather than a salary band.
• The M7 vice president, M8 senior vice president and C-suite rows are carried in the leadership section only. Including them in the band table would have forced every offshore market's factor toward the anchor at the top of the ladder and produced impossible local cells.

Peer-level mapping

Office and market catalogue — calibration factors versus San Francisco

Databricks confirms 30-plus offices in more than 20 countries but does not rank them by headcount, so this catalogue makes no claim about office size. Two markets in the source bundle are deliberately excluded here: Auckland, which the official office list does not show and which has no defensible band data, and Dubai, which appears as a hiring market in current careers listings but has no listed office address and no pay observation.

San Francisco anchor medians — the basis of every modeled cell

• The San Francisco column is the anchor. L3 through L7 medians come from the verified United States software-engineer aggregate, and M2 through M5 from the matching engineering-manager aggregate.
• Only ten of the nineteen markets carry a direct observation: San Francisco, Mountain View, Seattle and Bellevue, New York, Bengaluru, Amsterdam, Berlin, Belgrade, Toronto and Vancouver. Denver, Washington, Munich and Melbourne are office-confirmed but entirely modeled.
• Bengaluru is calibrated from direct L3, L4 and L5 observations at 26, 27 and 36 percent of the matched San Francisco base and 28.1, 24.8 and 27.3 percent of total compensation.
• Amsterdam is the best-evidenced non-United States ladder, with L3, L4 and L6 observations at 58, 52 and 47 percent of the matched United States total.
• London, Singapore and Sydney are calibrated from solution-architect and sales observations rather than software-engineering ones, so their factors describe field roles better than they describe an engineering ladder.

Executive incentive targets — percent of salary

• The equity pool size, the pool utilisation rate and the fully diluted share count are all unpublished, so no dilution or ownership-percentage calculation is possible from public data.
• Reported annual stock values cannot be converted into share counts. That conversion needs a grant price, a share class, a vesting cohort and a valuation methodology, none of which Databricks publishes.
• Nasdaq Private Market displayed an indicative $252.76 per share on 12 August 2026, with a $234 last trade and a $188.50 lowest offer. Those are third-party marketplace indications, not a company 409A value or a grant price, and they are deliberately excluded from every calculation in this report.
• Because there is no public exchange, a restricted stock unit that has vested is not the same as money. Value is realised only inside an announced tender window or at an exit, and neither is promised in any grant record that has been made public.

Valuation path and private-market reference points

Vesting — common reported employee schedule

The main reported schedule is unusually front-loaded: 40 percent of a grant vests in the first year and only 10 percent in the fourth, which is the reverse of the back-loaded schedules common in the private market and materially changes what a four-year offer is worth if the holder leaves early. An alternate even 25 percent a year schedule appears in the same dataset and may apply to different cohorts. The sources also conflict on cadence, with one showing monthly vesting inside each year and another describing immediate quarterly vesting with no cliff for some front-loaded new-hire grants. Only the individual grant agreement settles which applies.

Eligibility by hierarchy level

• Equity is broad rather than senior-only. Annual stock value appears at the entry level in all three countries with usable data, at $81,788 in the United States, 2.78 million rupees in India and €36,500 in the Netherlands.
• The equity share of the package rises steeply with level: annualised stock is 0.56 times base at L3, 1.10 at L4, 2.03 at L5, 3.84 at L6 and 5.24 at L7. From L4 upward stock is the larger half of the package.
• India equity is not a token. Annualised stock is 42.1 percent of reported total compensation at L3, 46.5 percent at L4 and 52.7 percent at L5, so a cash-only cost-to-company comparison understates an Indian offer badly.
• The management ladder is equity-heavy too, at 0.74 times base at M2, 2.67 at M3 and 3.51 at M4, so the M2 to M3 step adds roughly $464,000 of annualised stock on about $6,800 of extra base.
• There is no published refresh policy, no published grant matrix by band or country and no confirmation that every employee receives a grant. Refresh cadence should be negotiated explicitly rather than assumed.

Indicative annual grant value by band — San Francisco

Annualized planning value (±25%), not the face value of every new-hire grant. Highlighted rows are disclosed grant-date stock awards.

Named executive target equity

Two of these appointments were made by press release, Dave Conte as chief financial officer and Amy Reichanadter as chief people officer, and in both cases the release named the role and the background but no compensation arrangement, no new-hire equity value and no severance or change-of-control terms. That is the normal pattern for a private issuer and it means the arrival of a chief financial officer with public-company experience is the only outward signal about how the executive reward architecture may be evolving.

Board compensation framework

Chairman sitting fees, board equity and any lead independent director arrangement are all absent from the public record. This is a genuine disclosure gap created by private status, not a gap in the research: there is no filing in which the information would appear.

Not publicly disclosed

Modeled promotion planning timeline

The directional timeline that the community reports is roughly one and a half to three years from L3 to L4, two to four years from L4 to L5, three to five or more years from L5 to L6, and a highly selective step from L6 to L7 with no published timing at all. Manager progression from M2 to M5 is described as role and scope dependent. None of these is a Databricks statement, and the India dataset in particular shows total years of experience at a level rather than tenure in that level, so the two are easy to confuse.

Dataset summary

City-level H-1B wage history

All-years aggregates. P25/P75 are rounded reconstruction values marked modeled; medians and record counts are the stronger fields.

Selected title / worksite records

Evidence classification

Explicit “Not publicly disclosed” index

Known gaps and diligence before relying on a cell

1. 1The M7 vice president, M8 senior vice president and C-suite rows exist in the public level catalogue but carry no defensible pay observation whatsoever, so they are excluded from the band table and carried in the leadership section instead. Including them would have forced every offshore market's calibration factor toward the anchor at the top of the ladder and produced impossible local cells.
2. 2The M6 senior director row is base-only, from an official Zone 1 job posting. Its equity and bonus show zero because nothing is disclosed, not because nothing is granted, and the row is excluded from the range chart for that reason. The intern row is excluded on the same basis, being an annualised hourly rate rather than a salary.
3. 3The M4 aggregate publishes base and stock but no bonus, leaving a $36,000 gap to the reported $1.15 million total. That residual is carried as unallocated other compensation rather than being assigned to a bonus that no source reports.
4. 4The M5 director row uses the single Seattle submission because it is the only public component split at director level. The broader United States director aggregate shows a $1.20 million total median but publishes no components, so the row is one person's package rather than a market median, and its zero bonus is that person's outcome rather than a policy.
5. 5Two markets in the source bundle are excluded from the location table. Auckland does not appear on the official office list and has no band data, and Dubai appears as a current hiring market but has no listed office address and no pay observation. Both are named in the office note instead of being given invented factors.
6. 6Only ten of the nineteen markets carry a direct observation. Denver, Washington, Munich and Melbourne are office-confirmed with no pay data at all, and London, Singapore, Sydney, Tokyo and São Paulo are calibrated from field or sales observations rather than from a software-engineering ladder, so their engineering cells are a grade less reliable.
7. 7The two headline FY2026 H-1B counts disagree at 465 and 174 because the aggregators use different refresh dates, status filters and record definitions. Both are recorded rather than averaged, and no single count is adopted.
8. 8The vesting sources conflict. One shows monthly cadence inside each year of a 40/30/20/10 schedule, another describes immediate quarterly vesting with no cliff for some front-loaded new-hire grants, and a third schedule vests 25 percent a year. Only the individual grant agreement resolves which applies.
9. 9No total compensation cell in this report carries a notional employer benefit load. Every total is base plus reported bonus plus annualised equity and reconciles to its components; where a source published a benefit-loaded total reward figure, it has been rebuilt from the components instead.
10. 10Databricks publishes no geographic revenue split, so this report carries no regional revenue panel rather than repurposing one. Revenue is reported only as a company-wide run-rate.

FX rates used — 1 USD equals, snapshot 2026-08-25

Single-date conversion layer for comparability; it ignores payroll-date FX, tax, purchasing power, benefits valuation and hedging. Compensation intelligence, not legal, tax, investment, immigration or employment advice.

Source register — 19 sources

Headcount grew by roughly a quarter in about eighteen months on the company's own figures, but Databricks publishes no attrition rate, no headcount by city or country and no workforce compensation table, because a private issuer files no annual report containing them. The only geographic signals are the announced India target and the statement that the South Bay research team had doubled over two years and was projected to double again.