You have seen the list. AI Engineer, Data Science, Cyber Security, Data Engineering, Senior Software Engineer, FinTech, MLOps, DevOps, Chip Design, AI Product Manager, Forward Deployed Engineer, Robotics, Quantum, XR, Climate Tech. Fifteen job titles, presented as the safe bets for the next five years.
It is a reasonable list. Most of those roles are genuinely growing. But almost every version of it has two problems: the salary figures are American, and nobody checks the list against what employers are actually advertising.
So we did. Below is what UK employers paid, by job title, in the six months to 15 October 2026 — and the picture is messier than the list suggests.
Three of them are going backwards in the UK
Here is the uncomfortable part. Of the roles on that list with enough UK posting volume to measure, the two most prestigious — Software Engineer and Machine Learning Engineer — both show falling median advertised salaries. Meanwhile Network Engineer, the role that these lists routinely file under “avoid”, is up.
| Role | UK median | Year on year | Postings |
|---|---|---|---|
| Data ScientistPostings up from 140 | £80,000 | +18.5% | 682 |
| Data EngineerHigh, steady volume | £70,000 | +1.8% | 2,118 |
| DevOps EngineerHigh, steady volume | £70,000 | +1.5% | 1,532 |
| Software EngineerLargest sample on the list | £70,000 | −12.5% | 6,140 |
| Machine Learning EngineerSample too small to trust | £65,000 | −23.5% | 92 |
| Security EngineerRank up from 330 to 208 | £60,000 | +3.2% | 1,074 |
| Network EngineerThe role lists tell you to avoid | £57,500 | +4.6% | 1,271 |
Source: ITJobsWatch, UK permanent job vacancies, six months to 15 October 2026. Figures are median advertised salaries.
Before anyone screenshots that Machine Learning Engineer figure: it is built on 92 job postings. That is a thin enough sample that a handful of unusual adverts can move the median by twenty percent in either direction. Treat it as noise, not a trend.
The Software Engineer figure is different. That one rests on 6,140 postings — the largest sample on the list — and a 12.5% fall in median advertised salary, from £80,000 to £70,000, is a real signal. It is the single most important number in this article, and it is the opposite of what the list implies.
Worth knowing
“AI Engineer” — the number one entry on most versions of the list — is not a tracked job title on ITJobsWatch at all. Not low-paid, not declining: absent. Yet LinkedIn ranks it the fastest-growing job title in the UK.
Both are true, and the contradiction is the point. LinkedIn measures growth rate, which is flattering when you start from almost nothing. ITJobsWatch measures posting volume, and on that measure the title is still too rare in UK adverts to rank. A job can be the fastest-growing in the country and still be one you struggle to find.
What the American data says, and why it only takes you so far
The salary figures circulating with these lists almost always come from the US Bureau of Labor Statistics. The BLS is excellent, and its 2025–2035 projections are worth reading — but a Californian median salary tells a Manchester developer very little.
What the BLS projections are good for is direction of travel. And there, one comparison does more work than the entire top-15 list:
- Data scientists: projected to grow 35% between 2025 and 2035
- Information security analysts: 21%
- Software developers, QA analysts and testers: 10%
- Computer programmers: −7%
- All occupations, for comparison: 3%
Software developers and computer programmers sound like the same job to most people outside the industry. The BLS projects one growing three times faster than the average occupation and the other losing 8,100 jobs over the decade — and it attributes that decline directly to the automation of programming tasks and to higher-skilled work shifting towards developers.
That is the real lesson hiding inside the list. The dividing line is not which technology you work with. It is whether your job is specifying work or executing it.
The pattern that actually predicts who is safe
Once you line up the independent studies, the same split appears in all of them — and it is a seniority split, not a job-title split.
An analysis of 180 million job postings between January 2023 and October 2025 found total postings down 8% in 2025, but that decline was not spread evenly. Senior leadership roles fell 1.7%. Individual contributor roles fell 9%. The roles that collapsed were creative and administrative execution jobs — computer graphic artists down 33%, writers and photographers both down 28% — while creative direction roles held steady.
Indeed’s Hiring Lab found the same shape from the other direction. Software development postings grew almost 15% between February 2025 and June 2026, while overall postings fell 7%. But 71% of that increase came from senior roles, and 37% of it from jobs with AI in the title. Software roles overall remain 27.5% below their pre-pandemic level.
The short version
Demand did not disappear. It moved up the seniority ladder. The jobs being automated are the ones where somebody else decides what needs doing.
Which makes the entry-level picture the real story
If demand is concentrating at the senior end, the obvious question is how anyone becomes senior. The numbers there are genuinely bleak.
SignalFire’s 2025 talent report found new graduates now make up just 7% of hires at large tech companies — down 25% from 2023, and more than 50% below 2019. At startups, new grads are under 6% of hires. The share of new graduates landing roles at Alphabet, Amazon, Apple, Meta, Microsoft, NVIDIA or Tesla has more than halved since 2022.
The report names the trap precisely: companies post junior roles and then fill them with experienced people. You need prior experience to get the job that exists to give you experience.
That is a far more useful thing to know than which of fifteen job titles pays best. If you are early in your career, the binding constraint is not picking the right specialism — it is getting through the door at all.
What UK tech actually pays
Advertised salaries are one view, and a flattering one: they skew towards specialist, senior and London-weighted roles. For what people are genuinely paid, the Office for National Statistics surveys actual employees.
Here is every UK region for Information Technology Professionals — full-time, gross annual pay, 2025:
| Region | Lower quartile | Median | Top 10% |
|---|---|---|---|
| London | £51,963 | £70,000 | £117,622 |
| South East | £43,346 | £58,157 | £89,562 |
| East of England | £44,211 | £55,852 | £86,012 |
| South West | £42,154 | £54,150 | £83,391 |
| West Midlands | £40,912 | £53,320 | £82,113 |
| North East | £42,954 | £53,152 | £81,854 |
| North West | £40,297 | £51,394 | £79,147 |
| Yorkshire | £37,335 | £50,876 | £78,349 |
| Scotland | £39,768 | £49,665 | £76,484 |
| East Midlands | £40,756 | £49,599 | £76,382 |
| Wales | £37,954 | £48,344 | £74,450 |
Source: ONS ASHE 2025 (provisional), Table 15(3), SOC 2020 minor group 213.
Outside London, the median IT professional earns somewhere between £48,000 and £58,000. London’s median of £70,000 is 45% above Wales, and a top-decile London IT professional on £117,622 earns more than double the Welsh median.
That is the number to hold in your head when a video tells you a job pays £200,000. Somebody is earning that. In the UK, they are in the thin top slice of the top region — and the gap between the headline figure and the median is mostly geography and seniority, not job title.
So what should you actually do?
Do not pick a specialism off a list. Data Scientist is up 18.5% this year with postings rising from 140 to 682 — the strongest signal in the UK table. That is genuinely encouraging. It is also a snapshot of one year, and Machine Learning Engineer looked exactly that good twelve months ago.
Weight volume over headline pay. Software Engineer had 6,140 UK postings. Machine Learning Engineer had 92. A slightly lower median across thousands of openings is worth considerably more to your career than a high median across dozens, because the second one can vanish without anybody noticing.
Move towards specifying, away from executing. Every dataset here says the same thing. The work being automated is the work where the thinking has already been done by somebody else.
Check what the region pays, not what the internet pays. A £70,000 London offer and a £55,000 Manchester offer are not the comparison they look like, before you have even reached tax.
A note on these numbers
Advertised salary medians move with whatever employers happened to post that quarter, and low-volume job titles are volatile. We have flagged sample sizes throughout for that reason. Figures were accurate at the time of writing and are not career advice — they are a starting point for your own research.
Sources
- ITJobsWatch — UK permanent vacancies, six months to 15 October 2026
- ONS Annual Survey of Hours and Earnings 2025 (provisional), Table 15(3)
- US Bureau of Labor Statistics — Occupational Outlook Handbook, 2025–2035 projections
- Bloomberry — analysis of 180 million job postings, January 2023 to October 2025
- Indeed Hiring Lab — AI and job postings, July 2026
- SignalFire — State of Talent Report 2025
- World Economic Forum — Future of Jobs Report 2025