Secure Cloud environments for a national surveillance platform
Designed and stood up separated production, training, and testing cloud environments for a national surveillance platform and the access model that controls who can reach them.
Dec 2025 – ongoing
A national health surveillance platform needed more than one environment: somewhere live to run the real thing, somewhere safe to train staff, and somewhere to validate changes before they reached production. It also needed a clear answer to a question that usually goes unasked until something goes wrong; who can get into the live system, how, and with what powers.
Designed and stood up three separated environments — production, training, and UAT, each isolated so training activity or an untested change can never reach the live system, with monitoring and alerting on production so problems surface before users notice them. Then designed the access model: SSH removed entirely in favour of Session Manager, so no server is publicly reachable and every administrative session is brokered and accounted for, with least-privilege IAM groups separating infrastructure visibility from operational access and scoped to explicitly approved instances. Every permission set was validated in a test account before it reached production. When console access was refused while the command line worked, I traced it through policy simulation and session diagnostics to a missing console-specific read permission and added exactly that, nothing wider. All of it documented, so onboarding the next operations person is a procedure rather than an improvisation.
A production environment that can be administered without any server being publicly reachable, by people holding only the permissions their role actually requires with production and training responsibilities cleanly separated. The access model is documented and reusable, so onboarding another operations person is a known procedure rather than an improvised one.
National surveillance and IT teams.