Government
Building FinOps and observability on a government cloud
I owned the Observability and FinOps products on a secure, multi-tenant cloud platform used by UK law enforcement.
- Cloud platform
- AWS
- Azure
- Observability
- FinOps
- DevOps
- Governance
- AI
- £2.4Mannualised cloud savings
- 5% to 70%tagging compliance
- 10+teams on one Alert Manager
- 4.3/5AWS Landing Zone assessment
Overview
The Law Enforcement Cloud Platform is a secure, multi-tenant cloud used by UK law enforcement agencies. I owned its Observability and FinOps products, turning fragmented, under-owned capabilities into roadmaps focused on scalability, security, resilience and cost control.
Platform and observability
The problem
Observability was fragmented. Some teams used Grafana, some used CloudWatch and some tenants ran their own instances, with no shared or scalable approach.
Alarms fired with no clear owner or record of their purpose, there was no reliable way to check platform health, and the monitoring team’s scope was too narrow to give end-to-end visibility.
What I delivered
- A 24-month roadmap to move off the legacy monitoring solution, closing the critical gaps first, and I led its delivery.
- Operational and Protective Monitoring merged into one team, so one group owns visibility end to end.
- Alert Manager, a shared alerting service that enriches alerts and routes them to ServiceNow, Jira and email. I gathered tenant requirements, phased the delivery and led the onboarding of every platform team and tenant. More than 10 now use it in place of their own solutions.
- Alert Manager extended into the central routing layer, with instant messaging and AWS Health events, Security Hub findings and cost anomalies landing in the accountable team’s Jira backlog.
- The legacy Monitoring-as-Code estate retired, and Amazon Managed Grafana and Prometheus migrated to CloudWatch, with a service wrapper and templates so teams define their own alarms and dashboards.
- AWS Health and Security Hub dashboards, reporting and alerting, including monthly KPI reports with team health scores, leaderboards and vulnerability counts, each with a clear owner.
- A Platform Health status page, which I designed and specified, covering current availability, live incidents and past outages.
- Every critical and high security finding resolved, with all changes assured through IT Health Checks and Protective Monitoring testing.
FinOps
The problem
No FinOps practice existed. Cloud spend was accelerating with minimal visibility, tenants were charged on t-shirt-size estimates and nobody owned financial decisions across teams.
Tagging compliance was 5% and the platform scored poorly in external assessments.
What I delivered
- The platform’s FinOps practice built from scratch: the team, the foundations and a roadmap that moved cost management from reactive reporting to a core platform product.
- Cloud spend cut by £200k a month, £2.4M annualised, through Savings Plans and Reserved Instances, Graviton migration, instance scheduling that cut non-production costs by 60%, removing unused resources and automated clean-ups with lifecycle policies. A further £150k a month of savings has been identified.
- Tagging compliance raised from 5% to 70% with one tagging policy, CI/CD pipeline checks, Terraform and change-board controls, compliance dashboards, team leaderboards and Claude skills that help teams fix gaps. Around 120k untagged backups and snapshots were cleared through a new retention policy.
- T-shirt sizing replaced with a usage-based recharging model. I defined the calculations and billable items so every pound traces to an owner, and briefed tenants and finance. The platform no longer absorbs millions in unrecovered cost.
- Cost governance through per-team and platform budgets, spend dashboards for teams and leadership, and automated anomaly detection tracked in Jira.
- Platform KPIs across reliability, cost control, security, efficiency, modernisation and sustainability, reported in governance packs.
- FinOps by design across the delivery lifecycle, including Claude on Amazon Bedrock with budgets, per-user consumption tracking, dashboards and alerting.
- FinOps maturity moved from pre-crawl to walk in external assessments, and the platform scored 4.3/5 in the AWS Landing Zone assessment. I was then asked to scale the practice across a portfolio of AWS and Azure programmes.
More case studies
Explore other projects across government, platforms, e-commerce and financial services.