Our work

Case studies & selected projects

A selection of our work across QA, game testing and cloud operations. Some engagements are covered by NDA, so we describe the work without naming the client.

Discuss your project →

QA as a Service · Game Testing

Certifying a mobile and console title across five storefronts

Functional, compatibility and submission testing for a studio shipping the same title to Apple Arcade, Steam, Xbox, PlayStation 5 and Google Play Games.

  • Game testing
  • Compatibility
  • Submission readiness
  • Manual QA
  • Automation
Client
Confidential (under NDA)
Platforms
Apple Arcade, Steam, Xbox, PlayStation 5, Google Play Games
Services
Manual QA, compatibility and submission testing
Type
QA as a Service

The challenge

The studio had built the game but had no test function of its own. Each storefront applies its own technical requirements, and a failed submission costs a release slot as well as the rework. Internally, testing was being done by the developers between features, which meant regressions were found by players rather than before release.

What we did

  • Reviewed the build and each platform's current submission requirements, and turned them into a checklist per storefront rather than one generic pass.
  • Set up a device and console matrix covering the hardware the title had to support, including older mobile devices where performance problems usually appear first.
  • Ran functional and exploratory passes against the game's core loops, saving states and purchase flows, with reproducible steps and captured video for every defect.
  • Added long soak and stress sessions to surface memory growth and stability problems that only appear after extended play.
  • Automated the repeatable regression checks so each new build was verified before a manual pass began.
  • Produced a submission-readiness report per storefront, ranked by what would block certification versus what was cosmetic.

The outcome

The studio went into submission with a ranked, reproducible defect list and a per-platform checklist instead of a general impression that the game worked. Regressions were caught against each build rather than after release, and the developers stopped spending their own sprint time on manual testing.

DevOps & CloudOps · Google Cloud

A containerised platform on Google Cloud, managed as code

Replacing hand-built environments and manual releases with Kubernetes, infrastructure as code and an automated pipeline on Google Cloud.

  • Google Cloud
  • Kubernetes
  • Infrastructure as code
  • CI/CD
  • Monitoring
Client
Confidential (under NDA)
Platform
Google Cloud Platform (GCP)
Services
Containerisation, CI/CD, infrastructure as code, monitoring
Type
DevOps & CloudOps as a Service

The challenge

The client's services ran on Google Cloud, but the environments had been built by hand and every release depended on a sequence of manual steps known to one engineer. Deployments were slow to repeat, impossible to audit, and risky whenever that person was unavailable. Scaling meant provisioning more machines by hand, and nobody could say with confidence that staging matched production.

What we did

  • Assessed the existing services, environments and release process together with the client's developers, and agreed a target architecture before changing anything.
  • Packaged each service as a container so builds behave identically from a developer laptop through to production.
  • Deployed the services to a managed Kubernetes cluster on Google Cloud, with resource limits and autoscaling set against observed load rather than guesswork.
  • Described the cloud resources as code, so environments can be reviewed in a pull request and recreated consistently instead of being assembled by hand.
  • Built a CI/CD pipeline that tests, builds and deploys each change automatically, with a pre-release check and a documented rollback path.
  • Added monitoring, dashboards and alerting with runbooks, so an on-call engineer can act on an alert without waking a specialist.

The outcome

Releases moved from a manual, single-person process to an automated pipeline any engineer on the team can run. Environments are now recreated from code rather than memory, infrastructure changes are reviewed like any other change, and the platform scales with demand instead of requiring someone to provision machines.

Let's talk

Want something similar?

Tell us what you are trying to ship, or where delivery is getting stuck. A first conversation is free and there is no commitment.