Blog: Accelerating Research with a Scalable Cloud Platform at Anglia Ruskin University by Pamela Eteng, Cloud Platform Engineer
Context
Earlier this year I had the opportunity to work with Anglia Ruskin University (ARU) on a project focused on enabling research at scale in the cloud. ARU had recently been honoured with the Queen's Anniversary Prize for Higher and Further Education, a prestigious national award recognising outstanding achievement. As part of this recognition, the university secured a significant allocation of AWS research credits, providing the resources to pursue ambitious programmes powered by the cloud. Their ambition was to harness advanced machine learning and AI services such as Amazon SageMaker to support initiatives ranging from cancer detection through to the identification of misinformation online.
To turn this ambition into reality, ARU needed more than just access to cloud services. Each project required its own isolated account, with the right tools ready to use from day one. At the same time, the university needed a way to ensure that AWS credits were consumed responsibly. Automated budget enforcement was a critical requirement, alongside strong security and compliance guardrails. The goal was not simply to support one study, but to put in place a foundation for dozens of concurrent research initiatives—scalable and sustainable over the long term.
I joined the project as a cloud platform engineer, responsible for designing and deploying a landing zone aligned to best practices, and for building a budget enforcement solution that struck the right balance between researcher autonomy and institutional governance.
The process began with a workshop involving ARU's research stakeholders. These conversations were invaluable in clarifying what researchers really needed. Four themes consistently came to the surface: security, budget control, ease of access, and scalability. By anchoring my work around those priorities, I was able to keep the focus firmly on outcomes rather than technology for its own sake.
A key principle throughout was the use of Infrastructure as Code. I relied on AWS CloudFormation wherever possible to ensure everything was repeatable and consistent. This not only accelerated delivery but also meant the platform could expand seamlessly as more projects were added. At the heart of the design was a landing zone—a pre-configured cloud environment with built-in security, compliance, and account management guardrails—ensuring every research project started from a safe, consistent baseline.
Technical Solution
The platform took the form of a multi-account landing zone designed specifically for research. Each project was provisioned in its own AWS account to guarantee isolation and independence. On top of this, I built an automated budget enforcement system. This combined AWS Budgets, AWS Lambda, and Amazon SNS to provide automated enforcement of spend thresholds.
When a project neared its budget threshold, the system triggered automatically: stopping all SageMaker resources in the account and applying a Service Control Policy (SCP) to block access and prevent further spend. Researchers had the freedom to experiment with confidence while the process remained invisible until the point of enforcement, at which stage it acted decisively to prevent overspend.
The design was intentionally future-proofed. Centralised functions managed all accounts dynamically, while CloudFormation templates ensured that new projects could be provisioned quickly with the same controls in place. This combination of repeatability and automation meant that scaling to dozens of accounts was straightforward.
Personally, the most interesting challenge was making the enforcement system dynamic. I wanted it to handle multiple accounts automatically, inject SCPs seamlessly, and trigger only when necessary. Achieving that level of invisibility while still maintaining strict governance was a rewarding technical milestone.
Outcomes
Once live, the platform established a strong foundation for a new way of supporting research at ARU. Researchers could request a new project account and know that it would be provisioned quickly, with secure guardrails, budget thresholds, and access to SageMaker ready to go.
For individual researchers, the benefit was clear. Instead of waiting weeks for infrastructure to be provisioned or worrying about the risk of running up unexpected costs, they could begin work almost immediately in an environment already set up with the right tools, security, and budget protections. This meant more time focused on experiments and data, and less time navigating IT processes.
The risk of overspend was eliminated. With automated enforcement in place, the university could allocate its AWS credits with confidence, knowing that controls were consistent across every project. At the same time, researchers benefitted from autonomy—they had the freedom to experiment with ML and AI services without worrying about accidentally exceeding budgets.
Centralised logging and auditing made the platform easier to manage from an operational perspective, while the use of repeatable CloudFormation templates ensured that every new project was onboarded in the same controlled way. Reliability and compliance were built into the process, rather than bolted on afterwards.
Looking Ahead
The work completed at ARU represents the first stage of a longer journey. What has been delivered is a robust foundation—a platform that already supports multiple research initiatives and is designed to scale to many more. In time, the enforcement system can be extended beyond SageMaker to cover additional AWS services, creating even broader protection across research workloads. The platform also has potential to serve as a blueprint for other universities, offering a proven model that balances autonomy and governance in cloud-based research environments.
For me personally, the project was a reminder that the best solutions are often the simplest. By combining a strong architectural foundation with a handful of carefully chosen AWS services, we were able to meet complex requirements without unnecessary complexity. It was rewarding to see researchers able to move faster, scale their projects, and innovate with confidence—knowing that the guardrails were there to keep everything on track.