Insights
What AI work pays in the UK
Advertised day rates and salaries for 9 AI job titles, taken from published figures rather than from a survey we ran on ourselves. Every number shows the sample it came from. Where nothing is published, this guide says so instead of estimating.
The finding nobody else prints
Five of our 14 disciplines have no published rate in the UK advertised market at all. Four of those five are the governance, risk, safety and evaluation disciplines: the exact work every board is now being told it needs.
That absence is not a gap in the research. It is the finding. The advertised market has not yet learned to name this work as its own job, so it is being bought under titles borrowed from compliance, audit and legal. If you cannot find the salary, that is why. And it is why a job board cannot help you hire one.
- AI strategyNothing is published for this work as its own title. Strategy is advertised as a director or head-of role rather than a discipline, and those advertisements almost never quote a figure. Anyone who gives you a day rate for AI strategy is quoting their own book.
- AI governanceNo published rate or salary for AI governance as a title. The source returns governance roles that are IT governance, which is a different job. The work is being bought, but under titles borrowed from compliance and legal, so it does not appear in the data at all.
- AI risk and complianceNothing published. The work is real and rising with the EU AI Act timetable, and it is still being advertised as risk, audit or regulatory affairs with AI mentioned in the body. Until that changes there is no honest number to give you.
- Safety and evaluationAI safety does not exist as a title in the UK advertised market. What little there is sits inside research posts at a handful of labs and is not advertised with a figure.
- AI product managementNothing published for AI product management as a distinct title. Product roles that are entirely AI are advertised as product roles, so their figures are indistinguishable from product work generally.
- Training and enablementNothing published. Enablement work is bought as consultancy or training days, priced per engagement rather than per day, and none of it reaches a job board.
What a fractional engagement costs
Fractional work is priced from a day rate, not from a salary divided down. The two answer different questions: a salary buys continuity, a day rate buys days. The arithmetic below is the advertised middle for AI contract work, £551 a day, multiplied by a commitment. Nothing else.
| Commitment | Days a month | Per month | Per year |
|---|---|---|---|
| One day a week | 4 | £2,204 | £26,448 |
| Two days a week | 8 | £4,408 | £52,896 |
| Three days a week | 12 | £6,612 | £79,344 |
£551 a day multiplied by the commitment. That is the whole calculation. Nothing is uplifted for seniority, because any such uplift would be invented.
By discipline
Each page states whether the figures come from a matching title, from a broader one standing in, or from nothing at all.
- AI strategynothing published
- AI engineeringproxy
- LLM applicationsproxy
- Agents and retrievalproxy
- MLOps and platformdirect
- Data engineering for AIdirect
- AI governancenothing published
- AI risk and compliancenothing published
- Safety and evaluationnothing published
- Red teamingproxy
- AI product managementnothing published
- Fine-tuning and adaptationproxy
- Training and enablementnothing published
- Applied researchproxy
Every reading
Artificial Intelligence
as the source names it| Basis | Lower (10th) | Middle | Upper (90th) | Sample | Weight |
|---|---|---|---|---|---|
| Contract, per day | £400 | £551 | £788 | 2,974 rates quoted | strong |
| Permanent, per year | £40,000 | £71,000 | £111,250 | 5,017 salaries quoted | strong |
The broadest reading available. It covers every advertisement mentioning AI, so it spans research posts and jobs that mention AI in passing. Useful as a floor, useless as a target.
Outside London the permanent middle is £65,000, against £71,000 UK-wide.
Machine Learning Engineer
as the source names it| Basis | Lower (10th) | Middle | Upper (90th) | Sample | Weight |
|---|---|---|---|---|---|
| Contract, per day | £500 | £575 | £738 | 63 rates quoted | thin |
| Permanent, per year | £48,750 | £76,000 | £111,250 | 36 salaries quoted | thin |
A small sample on both sides. Thirty-six quoted salaries is not enough to read a percentile confidently, and the permanent median fell year on year on that thin sample rather than because the job got cheaper.
Outside London the permanent middle is £65,000, against £76,000 UK-wide.
Data Scientist
as the source names it| Basis | Lower (10th) | Middle | Upper (90th) | Sample | Weight |
|---|---|---|---|---|---|
| Contract, per day | £425 | £600 | £788 | 225 rates quoted | strong |
| Permanent, per year | £50,000 | £75,000 | £100,000 | 372 salaries quoted | strong |
Data Engineer
as the source names it| Basis | Lower (10th) | Middle | Upper (90th) | Sample | Weight |
|---|---|---|---|---|---|
| Contract, per day | £381 | £500 | £650 | 1,332 rates quoted | strong |
| Permanent, per year | £45,000 | £70,000 | £95,000 | 1,445 salaries quoted | strong |
The largest sample in this guide, and the only one where the percentiles are worth reading closely.
MLOps
as the source names it| Basis | Lower (10th) | Middle | Upper (90th) | Sample | Weight |
|---|---|---|---|---|---|
| Contract, per day | £450 | £575 | £790 | 239 rates quoted | strong |
| Permanent, per year | £56,750 | £87,500 | £118,750 | 196 salaries quoted | strong |
The highest permanent median of any reading here, and the highest tenth percentile by a wide margin. The floor is what is interesting: almost nobody advertises this work cheaply.
Generative AI
as the source names it| Basis | Lower (10th) | Middle | Upper (90th) | Sample | Weight |
|---|---|---|---|---|---|
| Contract, per day | £413 | £550 | £788 | 815 rates quoted | strong |
| Permanent, per year | Nothing published for this title. | ||||
Large Language Model
as the source names it| Basis | Lower (10th) | Middle | Upper (90th) | Sample | Weight |
|---|---|---|---|---|---|
| Contract, per day | not published | £550 | not published | not published | indicative |
| Permanent, per year | Nothing published for this title. | ||||
The source publishes a median for this term but no percentiles and no count of how many advertisements quoted a rate. A median with no spread and no sample size is a weak reading and is shown here as one.
Prompt Engineering
as the source names it| Basis | Lower (10th) | Middle | Upper (90th) | Sample | Weight |
|---|---|---|---|---|---|
| Contract, per day | not published | £550 | not published | not published | indicative |
| Permanent, per year | Nothing published for this title. | ||||
No percentiles published and no count of advertisements quoting a rate. Treat as an indication that the work is priced like other AI contract work, not as a rate you could plan around.
Red Teaming
as the source names it| Basis | Lower (10th) | Middle | Upper (90th) | Sample | Weight |
|---|---|---|---|---|---|
| Contract, per day | not published | £550 | not published | not published | indicative |
| Permanent, per year | Nothing published for this title. | ||||
Fifty-six advertisements in six months, no percentiles, and the term covers security red teaming far more often than model red teaming. The number is real and it is nearly meaningless for AI work specifically.
What this guide cannot tell you
- These are rates and salaries quoted in job advertisements, not rates and salaries paid. Advertised figures run ahead of agreed ones, and the gap widens at the top of a range.
- The figures are UK-wide. London and the South East sit above the national median for most technical roles, and the source publishes the outside-London figure separately where it differs.
- A day rate quoted to a contractor is not take-home. IR35 status, umbrella deductions and employer National Insurance can take a third or more of a rate determined inside IR35.
- A permanent salary excludes pension, bonus and equity. A day rate includes nothing at all: no holiday, no sick pay, no pension, and no paid time between engagements.
- Job titles are not a taxonomy. The source groups by the words used in advertisements, so a title like AI engineer covers work that ranges from prompt plumbing to distributed training.
- Where a sample is small, the percentiles move a lot between periods. Sample sizes are shown for exactly that reason.
Where the figures come from
Median and percentile day rates computed from rates quoted in UK contract job advertisements. The number of advertisements and the number quoting a rate are both published, so the sample behind each figure is inspectable.
6 months to 8 August 2026. Read 2026-08-08.
Median and percentile annual salaries computed from salaries quoted in UK permanent job advertisements, with the count of advertisements and the count quoting a salary published alongside.
6 months to 8 August 2026. Read 2026-08-08.
Guide updated 2026-08-08. Figures are taken again on 2026-11-08. If today is past that date and this line has not moved, the guide is stale and you should treat it as such.