Raj Prajapati

I designed and deployed a complete enterprise-grade AKS platform on Azure using Terraform IaC (modular ACR and AKS modules), nginx Ingress Controller with self-signed TLS certificate, Kubernetes manifests, and a GitHub Actions CI/CD pipeline that builds Docker images, pushes to Azure Container Registry, and deploys to AKS with rollout status verification. The platform runs two replicas with liveness and readiness probes and is managed via GitOps with ArgoCD and a self-healing policy.

A large enterprise operating in a regulated sector needed a container platform spanning both on-premises data centers and public cloud. Their existing estate was fragmented—legacy applications tied to bare metal and VMware, newer workloads pushing toward cloud, and no consistent deployment or operational model across either. Compliance requirements meant certain data classes could not leave on-premises infrastructure, so a cloud-only strategy was not viable. They needed a single platform experience regardless of where a workload actually ran.

My Role - I owned the platform architecture and led the deployment end-to-end—including infrastructure design, installation strategy, network and storage architecture, security model, and the operational handover that enabled their internal team to run it independently.

An enterprise data science team was running machine learning workloads on ad-hoc infrastructure — models deployed manually, no consistent serving layer, GPU capacity allocated by informal arrangement rather than policy, and no visibility into whether models were degrading in production. Inference costs were rising with no clear attribution; GPUs sat idle during off-peak periods while teams queued for capacity during busy periods; and there was no mechanism to detect when a model's input data had drifted from what it was trained on.

My Role — I designed and built the AI platform layer—GPU infrastructure, model-serving runtime, deployment pipeline, and observability that made both cost and model health visible.