Skip to content
Case study · Research · National

Secure data, a national service re-platformed, and the devices it all runs on.

Three strands of work for the body that funds the nation’s research. A secure research environment on AWS in which external researchers work with the funder’s data, build machine-learning models and layer on large language models, proven as a pilot and adopted as the standard model. Technology leadership and architecture oversight of the national research-assessment programme: a new architecture, a technical plan, engineering practices and a platform rebuilt on Azure, with the same team taking the submissions service to release on outcome-gated milestones. And an evidence-led plan for the organisation’s end-user devices, keeping the Windows 11 rollout moving while the future service model is decided.

The figures

What the engagement delivered, in numbers.

3 strands

A secure research environment, a national service rebuilt on a new architecture, and the end-user device estate: one client, one way of working.

Adopted as standard

The secure research environment became the funder’s standard model for safe external access to its data.

Rebuilt on Azure

Target architecture, technical plan and engineering practices in place, and the national service’s platform rebuilt.

13 milestones

Outcome-gated monthly milestones on a fixed fee to a release-ready service, load-tested for around 5,000 concurrent users.

How the engagement ran
01
Where they started

Three problems, one funder.

Sensitive data that external researchers needed to work with, without the funder giving up control of it. A national research-assessment programme, inherited across several clouds and identity mechanisms, that needed a single architecture and modern engineering practice before a timetable that does not move. And a Windows 11 rollout across around 2,000 devices, with two candidate builds and a decision to make on how the device service should be governed and delivered.

What this shows

What this shows

One client, three very different problems, one way of working: understand the estate, decide the order, build to standards, hand over capability.

AI can run on sensitive data without the data leaving the organisation’s control, when the models are hosted inside the same boundary as the data.

A fixed monthly fee with outcome-gated milestones puts delivery risk with Scout and gives a public body a plan it can defend to its board and its auditors.

“The timetable for the assessment exercise does not move, and every university in the country depends on the service. We needed the programme on a footing we could defend: an architecture, a plan, engineering practices, and a platform rebuilt properly. Scout gave us that with a small senior team, and now takes us to release on outcomes we can gate, with our own people able to run it without them.”

Programme Sponsor

Keep reading

Talk to us about your organisation.

Tell us where your organisation is starting from.