Valentine Anagbogu
I build the paved roads engineering teams ship on: self-service developer platforms, secure Azure foundations, and Kubernetes that stays up when it matters.
I'm a senior platform engineer. I design and run the foundations product teams build on: Azure landing zones, large shared AKS clusters, and an internal developer platform that turns a single form into a running, secured, monitored service.
My work sits where infrastructure, security and developer experience meet. I build golden paths with Backstage, Terraform and Helm. I keep networking private and CI/CD keyless by default. When production misbehaves, I lead the investigation: evidence first, theory second.
I care about making the right way the easy way, with standards in code, changes that are safe to make, and cost decisions backed by evidence. The case studies below show how I work.
Backstage golden paths for Node.js, .NET and Python: repo, keyless CI/CD, RBAC and deployment from one form.
Reliability, incident root-cause analysis and high-availability standards on multi-tenant AKS clusters with 200+ nodes.
Hub-and-spoke networking, private endpoints, firewall policy and private DNS, all as Terraform.
Cost reviews backed by live resource data, workload identity, and policy enforced in code.
CKA · Azure Solutions Architect Expert · Azure DevOps Engineer Expert · AWS Solutions Architect · Terraform Associate
CKA
2025
AZ-400
2023
AZ-305
2024
AZ-104
2022
SAA-C02
2022
CLF-C02
2022
002
2022
A self-service Backstage portal with golden paths for Node.js, .NET and Python. One form creates the repo, keyless CI/CD, RBAC, infrastructure and an AKS deployment.
View Project →Taking a proof-of-concept portal to production on shared AKS: workload identity, private endpoints, a dedicated node pool, group-gated SSO and six silent issues fixed along the way.
View Project →Intermittent production timeouts with no change on our side. The cause was a provider-managed CoreDNS update, made worse by a liveness probe wired to downstream health.
View Project →Batch Jobs in every namespace started minutes late. One replica of a sidecar-injection webhook was throttling pod creation. Scaling it and forcing reconnects brought starts down to seconds.
View Project →Hub-and-spoke networking with firewall policy as code, private-only PaaS and a single authoritative private DNS pattern that resolves correctly from AKS, VMs and on-premises.
View Project →An estate-wide cost review that validated every finding against live state. Always-on non-production AKS clusters were the biggest lever in a five-figure monthly saving.
View Project →A memory-saturated node took a single-replica auth service with it, and the autoscaler had no room to recover. This is the RCA and the high-availability baseline that came out of it.
View Project →Three services, a thousand restarts each, and one shared symptom: "health check failed". Testing layer by layer from inside the pod found three different private DNS root causes.
View Project →Users saw a block page, and the first instinct was to change the cluster. Testing from the pod outwards proved every layer healthy and found the block upstream, on the client side.
View Project →Five ways a shared Helm chart and pipeline reported success while the app was broken, from dropped args and unbound ports to sticky image tags, and the guardrails that stop them.
View Project →Terraform pipelines with no stored cloud secrets, plus the root cause of AADSTS700024 failures on long applies: a token exported once and never refreshed.
View Project →Rotating a shared TLS certificate across dozens of services without an outage, distributing SFTP host keys to every environment and triaging expiring app-registration secrets.
View Project →Three observability investigations: a health check flooding a log partition, a missing partition hiding millions of dev logs a day, and an averaged disk metric hiding a full volume.
View Project →Re-pointing diagnostics for a set of Terraform stacks to a new region's Log Analytics, one environment at a time, with a clean revert-and-reland and drift handled rather than overwritten.
View Project →Database platform changes in Terraform (renames, one-way settings, firewall-blocked Entra auth, decommissioning), each with the check that kept it safe, including a mass destroy caught at plan time.
View Project →Platform runbooks and golden-path standards encoded as AI agent skills. Investigations follow the same evidence trail every time, and resource requests become reviewed pull requests.
View Project →