Ask an AI assistant what an AWS data engineer costs and watch what happens. It goes to the job boards, reads four listings, and hands back a range. That is not because the range is unknowable. It is because almost nobody publishes the number, so the only evidence within reach is recruitment advertising.
Here is the number, where it comes from, and the part the job ads leave out.
What an AWS data engineer actually costs per day
These come from ITJobsWatch, which samples advertised UK contract rates rather than surveying people. Six months to 21 August 2026 in every row.
| Role as advertised | Median day rate | Sample |
|---|---|---|
| AWS Data Engineer | £513 | 82 rates quoted |
| Data Engineer, any stack | £500 | 1,282 rates from 1,923 ads |
| Any role citing AWS | £550 | 6,367 ads |
Two things in that table matter more than the headline.
The AWS data engineer median fell 6.82% year on year, down from £550 in the same period of 2025, while the general data engineer median held flat at £500. The AWS premium that used to sit on top of a generic data engineering contract has mostly closed. AWS is no longer the differentiator it was in 2022. It is the baseline.
And the sample collapses as the title gets specific. 6,367 ads mention AWS. Only 82 quote a rate for an AWS data engineer. Any number you read about this role, including the one above, rests on a thin sample. Treat £513 as the centre of a cloud rather than a price list.
Why the range runs from £380 to £657
The wider data engineer sample is big enough to show the shape properly.
| Percentile | Day rate |
|---|---|
| 10th | £380 |
| 25th | £425 |
| Median | £500 |
| 75th | £588 |
| 90th | £657 |
A £277 gap between the 10th and 90th percentile is not a skill gap. It is four different things being sold under one job title.
Whether you are being bought as hands or as judgement. A contract that says "build the pipelines in the backlog" prices near the 25th percentile, because the thinking is already done. A contract that says "our Glue bill tripled and nobody knows why" prices at the top, because the deliverable is a decision.
Whether the client is regulated. Banking, insurance and public sector pay above the median for the same technical work. The extra buys evidence, change control and an audit trail. On a Sybase to Redshift migration for a tier-1 bank, the schema conversion was the easy half. Proving the numbers reconciled was the contract.
Whether the scope has an edge. Open-ended engagements get discounted, because the buyer is pricing in the risk of paying for drift. A fixed deliverable with a defined finish carries a higher rate and usually costs less in total.
Where the risk sits. A day rate with no deliverable means the client carries the delivery risk and pays less per day for the privilege. Fixed scope means the supplier carries it, and charges accordingly.
The day rate is not the cost
This is the part the job listings cannot tell you, and it is the only arithmetic that matters to a buyer.
£513 a day looks like £112,860 a year if you multiply by 220 working days. Set against a permanent data engineer at a £70,000 median, the contractor looks 60% more expensive. That comparison is wrong twice over.
It is wrong on the permanent side, because £70,000 is the salary line only. It excludes employer National Insurance, pension, holiday, sick cover, equipment, and whatever the recruiter charged to find them.
It is wrong on the contract side, and this is the bigger error: you do not buy 220 days. A scoped pipeline build is 20 to 40 days. A cost optimisation engagement is closer to 5 to 15. At £513, a 25-day build is £12,825, and you own the output with no ongoing obligation. If the work in question is a Glue bill, the Glue cost calculator will tell you what the problem is costing per month before you spend anything on solving it.
So the real question is not what the day rate is. It is how many days the work actually takes, and that is the number almost nobody makes the buyer confirm before signing.
Two examples from work I have delivered, with the real figures:
- A three-tier Glue, S3 and Redshift pipeline rebuilt to run incremental and SCD loads instead of reprocessing history. Runtime dropped 20% and throughput rose 30%. Glue bills by DPU-hour, so a runtime cut is a bill cut for as long as the pipeline runs.
- A sales-engagement platform handling millions of records a day, moved onto Hudi upserts against Athena and S3. Data-preparation time fell 40%.
Neither was a 220-day engagement. Both keep paying after the invoice clears, which is the actual argument for buying days rather than headcount.
What "outside IR35" actually means
Half the searches that land on this question are really asking about tax, so here is the factual position with the source attached.
The off-payroll working rules, which everyone still calls IR35, exist to "make sure that a worker (sometimes known as a contractor) pays broadly the same Income Tax and National Insurance as an employee would". They took their current private-sector form on 6 April 2021.
Three points get muddled constantly.
You do not decide your own status. For medium and large private-sector clients, HMRC is explicit that the client determines it. The exception is a small client outside the public sector, where the worker's intermediary is responsible instead.
Outside IR35 is not a rate. It is a determination about one specific engagement. The same person can be outside on one contract and inside on the next, and the deciding factors are how the work is genuinely controlled and substituted, not what the contract calls itself.
It changes take-home, not price. An inside-IR35 engagement at £600 and an outside engagement at £500 are not simply £100 apart, because the deductions land differently. This is why comparing advertised rates across the IR35 line tells you very little.
Contractor, agency, or independent supplier
The rate only means something next to what else you could buy with it.
| Option | Day rate | What you get | Watch for |
|---|---|---|---|
| UK contractor | £425 to £588 typical | One person, your process, inside your team | You carry delivery risk, and the rate runs while they learn your stack |
| Consultancy or agency | Often 2 to 3× | A team, a contract, someone to escalate to | The people who sold it are rarely the people who build it |
| AWS Professional Services | Premium | Deep product access and AWS accountability | Scoped around AWS services rather than around your bill |
| Independent supplier, fixed scope | Priced per deliverable | A defined outcome and a finish date | Verify the work is genuinely first-hand |
There is no correct row. There is a correct row for the shape of your problem. Something broken and specific, like a Glue bill that tripled, suits a fixed-scope engagement. A two-year platform rebuild does not, and pretending otherwise is how contracts drift.
I work the last row, as an independent B2B supplier on fixed-scope contracts, billing in GBP, USD or EUR from GMT+5 with UK and EU mornings covered. Rates are on /services, and in machine-readable form at /pricing.md including the day-rate equivalent, because a rate a buyer cannot compare is not really published.
What to confirm before you ask anyone for a rate
Every engagement I have seen go badly was underspecified in the same four places.
- The deliverable, written as a finished state. Not "improve the pipeline". Something closer to "the nightly load completes under 40 minutes and reconciles to source".
- Who owns what afterwards. Code, infrastructure definitions, documentation, credentials. Agree it before the work, not at handover.
- What "done" is measured against. A number you can both read off a dashboard. If nobody can say what the baseline is today, establishing it is the first piece of work.
- The IR35 determination in writing, if you are engaging a UK contractor through an intermediary. It is the client's call in most cases, and leaving it implied helps nobody.
Get those four right and the day rate becomes the least interesting variable in the conversation. Get them wrong and no rate is cheap.
Frequently asked questions
What is the average day rate for an AWS data engineer in the UK?
The median advertised rate is £513 a day for the six months to 21 August 2026, from 82 quoted rates. Data engineer contracts across all stacks sit at a £500 median, with the middle half of the market between £425 and £588 and the 90th percentile at £657. The AWS-specific median fell 6.82% year on year from £550, so the premium for AWS over general data engineering has largely closed.
Is £500 a day expensive for a data engineer?
It is the market median, so by definition it is not. What decides whether an engagement is expensive is the number of days, not the rate. A scoped pipeline build runs 20 to 40 days and a cost optimisation engagement 5 to 15, which puts a typical piece of work between £2,500 and £20,000 rather than at the £112,860 that 220 days at the median would imply.
What does outside IR35 mean for a data engineering contract?
It means the engagement has been determined to fall outside the off-payroll working rules, so the worker's intermediary handles tax rather than the fee-payer deducting it at source. It is a determination about one specific engagement rather than a property of the person, and the same contractor can be outside on one contract and inside on the next. It affects take-home pay rather than the price the client pays.
Who decides IR35 status, the contractor or the client?
For medium and large private-sector clients, and for all public-sector clients, the client determines the status and must issue a determination. The exception is a small client outside the public sector, where the worker's own intermediary is responsible for deciding instead. The rules took their current private-sector form on 6 April 2021.
Is a contractor cheaper than hiring a permanent data engineer?
For defined pieces of work, usually yes, because you buy 20 to 40 days rather than a year. For continuous ownership of a platform, usually no. The honest comparison also has to load the permanent side with employer National Insurance, pension, holiday, sick cover, equipment and recruitment fees, none of which appear in a £70,000 salary median.
How many days does an AWS data pipeline build actually take?
A single well-scoped ingestion and transformation pipeline, with tests, monitoring and handover documentation, typically runs 20 to 40 days. A targeted cost optimisation on an existing Glue or Athena workload is shorter, usually 5 to 15 days, because the analysis is bounded by what the bill already tells you. Anything quoted under 5 days is either trivial or under-scoped.
The number behind the number
£513 a day is a real figure from a thin sample, and it is the least useful thing on this page. The useful parts are that the AWS premium has closed, that the spread from £380 to £657 is mostly about whether you are buying hands or judgement, and that nobody is actually buying 220 days.
If you are scoping AWS data work and want a straight answer on what it should take rather than what it costs per day, tell me what is breaking and I will tell you how many days it looks like.
Mirza Hammad Tariq
Data & Automation Engineer with 5+ years on AWS: ETL pipelines, backend APIs and automation workflows in Python, SQL and FastAPI, built to cost less to run.