Machine Learning Engineer salary in India, the UK and the US
What 2,512 job postings carrying a salary figure say this role pays — the whole distribution, the sample size behind each figure, the date it was measured, and how much of it the source estimated rather than read off an advertisement. Not a single national average with nothing under it.
Jump to: Indian · UK · US · Method
Indian salaries 153 postings with a figure · measured 2026-08-22
Across 153 machine learning engineer salary figures in India, the middle half runs from ₹9.8L to ₹24L, with the midpoint near ₹16L. The largest single band is ₹10L to ₹20L, holding 41% of postings.
In a sample of 250 Indian postings, 0% of salary figures were estimated rather than stated — these are close to entirely quoted from the advertisements themselves. Only 4% of Indian postings carry a figure at all, which is the bigger limitation here.
The full distribution
| Band | Postings | Share |
|---|---|---|
| ₹0L to ₹10L | 39 | 25.5% |
| ₹10L to ₹20L | 62 | 40.5% |
| ₹20L to ₹30L | 34 | 22.2% |
| ₹30L to ₹40L | 10 | 6.5% |
| ₹40L to ₹50L | 5 | 3.3% |
| ₹50L to ₹60L | 3 | 2.0% |
| ₹60L and above | 0 | 0.0% |
Who is posting this role in India
Ranked by how often each employer appears across machine learning engineer advertisements in the source index, most first. We show the order and not a count, because the underlying figure counts advertisements rather than vacancies: a staffing agency posting one role across forty cities outranks a company hiring four people, and bulk public-sector and franchise feeds replicate a single advertisement thousands of times. The order is informative; the number would not have meant what it appeared to mean.
| # | Employer |
|---|---|
| 1 | Oracle |
| 2 | HP Textile Mill |
| 3 | H & R Johnson |
| 4 | Maersk |
| 5 | Micron Electricals |
UK salaries 187 postings with a figure · measured 2026-08-22
Most machine learning engineer salary figures in the UK cluster at the top of the range: 73% of 187 postings sit at £70k or more. A median is not a useful summary of a distribution shaped like this — the useful figure is that share.
In a sample of 250 UK postings, 29% of the salary figures were estimated by Adzuna rather than stated in the advertisement. The rest are quoted from the posting.
The full distribution
| Band | Postings | Share |
|---|---|---|
| £10k to £20k | 0 | 0.0% |
| £20k to £30k | 0 | 0.0% |
| £30k to £40k | 4 | 2.1% |
| £40k to £50k | 4 | 2.1% |
| £50k to £60k | 11 | 5.9% |
| £60k to £70k | 32 | 17.1% |
| £70k and above | 136 | 72.7% |
The last 12 months
The monthly average has risen by 11.5% between August 2025 (£81,745) and July 2026 (£91,132). This is a monthly average across every posting that carried a figure, stated or estimated, so it moves with the mix of roles being advertised as well as with pay itself — a month heavy on senior contracts lifts it without anyone getting a rise.
Who is posting this role in the UK
Ranked by how often each employer appears across machine learning engineer advertisements in the source index, most first. We show the order and not a count, because the underlying figure counts advertisements rather than vacancies: a staffing agency posting one role across forty cities outranks a company hiring four people, and bulk public-sector and franchise feeds replicate a single advertisement thousands of times. The order is informative; the number would not have meant what it appeared to mean.
| # | Employer |
|---|---|
| 1 | Spotify |
| 2 | Bloomberg |
| 3 | Goldman Sachs |
| 4 | Amazon |
| 5 | JPMorgan Chase |
US salaries 2,172 postings with a figure · measured 2026-08-22
Most machine learning engineer salary figures in the US cluster at the top of the range: 65% of 2,172 postings sit at $140k or more. A median is not a useful summary of a distribution shaped like this — the useful figure is that share.
Most of these are estimates, not quotes: in a sample of 250 US postings, 70% of the salary figures were calculated by Adzuna from the posting rather than stated in it. Read the distribution as the source's model of the market, not as what employers wrote down.
The full distribution
| Band | Postings | Share |
|---|---|---|
| $20k to $40k | 0 | 0.0% |
| $40k to $60k | 31 | 1.4% |
| $60k to $80k | 78 | 3.6% |
| $80k to $100k | 393 | 18.1% |
| $100k to $120k | 120 | 5.5% |
| $120k to $140k | 130 | 6.0% |
| $140k and above | 1,420 | 65.4% |
The last 12 months
The monthly average has fallen by 5.0% between August 2025 ($182,494) and July 2026 ($173,284). This is a monthly average across every posting that carried a figure, stated or estimated, so it moves with the mix of roles being advertised as well as with pay itself — a month heavy on senior contracts lifts it without anyone getting a rise.
Who is posting this role in the US
Ranked by how often each employer appears across machine learning engineer advertisements in the source index, most first. We show the order and not a count, because the underlying figure counts advertisements rather than vacancies: a staffing agency posting one role across forty cities outranks a company hiring four people, and bulk public-sector and franchise feeds replicate a single advertisement thousands of times. The order is informative; the number would not have meant what it appeared to mean.
| # | Employer |
|---|---|
| 1 | Huntington Ingalls Industries |
| 2 | Oracle |
| 3 | Capital One |
| 4 | General Motors |
| 5 | PwC |
Using this in an actual negotiation
A number is only worth having if you can say it out loud under pressure. Three things make that easier.
Quote the range and its source. "Advertised machine learning engineer postings in India sit mostly between ₹9.8L and ₹24L" is a checkable claim. It moves the conversation onto evidence, which is a conversation you can win. A figure with no source behind it invites the reply "that doesn't match our banding", and you have nowhere to go.
Anchor high in the band, not outside it. If your experience is genuinely above the middle, name the upper quarter and give one concrete reason — a system you own, a number you moved, a scarce skill this posting names. Anchoring outside the observed range without evidence reads as unserious and costs you the credibility you need for the rest of the conversation.
Know what you are comparing. Base is not total, and total is not take-home. Run the offer through the take-home pay calculator before you decide anything, and if you are weighing two offers in different cities, the offer comparison calculator adjusts for cost of living so you are comparing like with like.
Then rehearse it. The negotiation scripts cover the four moments where most people give up money: the early "what are your expectations?", the first offer, the counter, and the exploding deadline. Reading them is not the same as saying them — the whole difficulty is that these sentences are uncomfortable to speak, and the only fix is having said them before.
What these numbers are, and what they are not
Source. Salary bands, monthly averages and employer counts come from the Adzuna jobs API, which aggregates published job advertisements. Figures were last refreshed on 2026-08-22; this page was built 2026-09-01.
Some of these are estimates, not quotes. Adzuna calculates a salary for postings that state none, and its histogram counts those alongside real ones. We sampled 250 postings across five roles in each market to measure the split: India 0%, the UK 29%, the US 70% of the salary figures are estimated rather than stated. Each country section repeats its own share. It was tempting to call all of this "advertised salaries" — the endpoint is named for salaries and the word reads well. It would have been false.
They are postings, not payroll. Not offers and not what anyone receives. Most postings never state pay at all, and the ones that do skew towards agency, contract and public-sector listings where quoting a band is standard practice. That bias generally pushes these figures away from what a typical permanent in-house role pays.
They are bands, not points. The source reports counts per salary band, so a quartile can only be located inside a band. Where this page prints a single figure it is a linear interpolation within the band that contains that quartile, and where a quartile falls in the open-ended top band it says "at least" instead. Nobody can compute a median to the rupee from bucketed data, and pages that print one are not measuring — they are decorating.
The query is stated. Each section reports the search that produced it — "machine learning engineer" here. A salary figure is only as specific as the query behind it, so if the query is broader than the job title you have in mind, the figure is too.
Sample size is always shown. No market appears on this page with fewer than 40 postings behind it. Where you see a section missing, that is why.
Found something wrong? Tell us — corrections are published, not quietly patched.
Questions
What does a machine learning engineer actually earn?
In India, the middle half of 153 machine learning engineer salary figures runs ₹9.8L to ₹24L. In the UK, the middle half of 187 machine learning engineer salary figures runs £69k to at least £70k. In the US, the middle half of 2,172 machine learning engineer salary figures runs $107k to at least $140k. These come from job postings rather than from a payroll survey, most postings carry no figure at all, and a share of the figures that exist are the source's estimate rather than a number the employer published — that share is stated per country on the page.
Is this the average machine learning engineer salary?
No, on two counts. It measures job postings rather than payroll, and postings that carry a figure are not a random sample of jobs — agency and contract listings quote pay far more often than in-house permanent roles, and they price differently. It also includes salaries the source estimated for postings that stated none: 0% of the Indian figures, 29% of the UK figures and 70% of the US ones. Read it as the posted market plus a model of it, which is still the market you negotiate against.
How do I use these numbers in a salary negotiation?
Quote a range, not a point, and say where it comes from. "Advertised postings for this role in my market sit mostly between X and Y" is checkable and hard to dismiss; "I think I'm worth Z" is not. Anchor at the upper quarter if your experience is genuinely above the middle of the band, and have one specific piece of evidence ready for why.
Why is the top band open-ended?
The source reports the highest band as "₹60L and above" with no ceiling, so any figure past that point would be invented. Where a quartile falls inside it, this page says "at least" rather than guessing a number.