I lead the Developer Experience team at Unifonic and set platform direction for a multi-cloud Kubernetes estate across GCP, AWS, OCI, and Azure. Lately that includes building the AI layer: MCP servers that give assistants real project and cluster context, and AI-assisted review that holds quality steady as code volume grows.
Each one covers the context, what I did, and the call I had to make.
Most AI assistants are guessing about your systems. I built the servers that let them stop guessing: real context from Jira, Confluence, Bitbucket, and live Kubernetes clusters, plus AI-assisted review that keeps quality steady as code volume grows.
Multi-cloud Kubernetes from CI/CD through observability, security, and DR.
Six engineers embedded two-to-a-squad instead of one central ticket queue.
A modernization program stalled 8+ months, delivered in 3 after the direction changed.
Grew DevOps from 3 engineers to 9 while revamping infrastructure for 500+ microservices.
The AI layer is live, the OCI migration is delivered, and the golden path is the target state in build.
An assistant that cannot see your tickets, your repositories, or your clusters is guessing. These servers give it the same context an engineer has, which is what makes its review output worth trusting.
Built in-house rather than bought, so the context layer matches how the team actually works.
Saudi data-residency rules meant customer data had to stay in country. The migration was staged rather than a lift and shift, so registry resilience and multi-region continuity existed before anything cut over. Production now runs on OCI.
OCI was chosen over staying on AWS because residency ruled out the alternative, and because it came in at under half the cost while still being a full cloud.
The goal is that creating a service is one action, not a ticket and six follow-ups. A template scaffolds the repository with delivery, observability, security, and policy already wired in, and the catalog knows who owns it.
Backstage is in build. The embedded team model that keeps these templates alive is already running.
If your team is modernizing cloud platforms, building an internal developer platform, or figuring out where AI actually fits in the toolchain, I would be glad to talk.