DevOps Engineer with 3.7 years building and running Kubernetes platforms — cluster provisioning, GitOps delivery, CI/CD, and observability. Across managed control planes (EKS, GKE), enterprise platforms (OpenShift, vSphere) and bare metal, driven end to end by Helm, Ansible and Terraform rather than clicked together in a console.

What I care about is the part most tutorials skip: what happens when the deploy reports success and the cluster is still broken. My repos ship the runbook and the root-cause writeup alongside the code, because the second one is what you actually need at 2am.

role:      DevOps / Platform Engineer
focus:     Kubernetes platform engineering, IaC, GitOps, DevSecOps, observability
approach:  version-pinned · idempotent · reproducible · documented
currently: deepening SRE practice — reliability, incident response, platform UX
open_to:   DevOps · Platform Engineering · SRE

Container platform & orchestration

Cloud & virtualization

Infrastructure as code

CI/CD & GitOps

Security & secrets

Observability & APM

AI engineering

Languages & scripting

Data

Systems

Container to Kubernetes with the security gates actually enforced, not just reported. GitHub Actions runs Trivy over filesystem and image, generates a CycloneDX SBOM, and gates the build on CRITICAL/HIGH. Conftest enforces 11 OPA/Rego rules against rendered manifests, and a deliberately insecure fixture in the repo proves those rules bite. Secrets come from Vault via the External Secrets Operator using Kubernetes auth — nothing sensitive in git. The image is a scratch base with a static Go binary, non-root, read-only rootfs, all capabilities dropped.

Hybrid Helm + Kustomize deployment of Kubeflow 26.03.1 on bare-metal Kubernetes — KServe, Kubeflow Trainer, Spark Operator, Istio and cert-manager, version-pinned and idempotent end to end. The kind of install that usually turns into a week of yak-shaving, reduced to a repeatable run.

Kubernetes cluster provisioned with Kubespray + Ansible, Make-driven and reproducible from a single config file. Ships with a full runbook and a root-cause analysis of a CoreDNS forwarding loop that left DNS dead while Ansible cheerfully reported failed=0 — the diagnosis, the fix, and why it belongs in the inventory rather than on the node.

Standalone Terraform configurations for core AWS services — VPC, EC2, ALB, RDS, ECR, Lambda, IAM, Secrets Manager, EventBridge and Step Functions. Composable building blocks rather than one monolithic stack.

Helm chart deploying KubeDoom with hardened, tightly scoped RBAC. A genuinely fun project that doubles as a practical exercise in least-privilege service accounts.

Step-by-step guide and command logs for a two-node MongoDB replica set on Ubuntu — written so someone else can follow it without guessing.

Working toward public release: Argo CD self-healing GitOps, Jenkins X on Ansible-provisioned HA Kubernetes, Velero backup and restore, a full Prometheus/Grafana/Loki monitoring stack, OpenTelemetry instrumentation, and Bazel-built CI pipelines.

rndudhe1808@gmail.com

_{Open to DevOps, Platform Engineering and SRE roles.}