Senior DevOps / Platform Engineer. I build and run the systems underneath — Kubernetes, GPU clusters, GitOps and service mesh — and keep them up while everyone else is asleep.
Five years. Azure and AWS. Clusters built from the ground up with Terraform. GPUs sliced for AI teams. Ingress rewritten as a mesh. Five Kubernetes upgrades with zero downtime. The users never noticed. That's the point.
Name: ajith-kumar Namespace: platform Labels: role=senior-devops focus=kubernetes,gpu,gitops certs=az-400,az-104,az-900 edu=btech-it,cgpa-7.95 Annotations: motto: "if it's not in git, it doesn't exist" Status: Running Node: commotion/azure-aks Containers: platform: Image: kubernetes:v1.34 Limits: nvidia.com/gpu: H200-NVL Events: Type Reason Age Message ---- ------ ---- ------- Normal Scheduled 5y started at Cognizant Normal Pulled 3y kovai.co — AKS, KEDA, Turbo360 Normal Started 1y senior devops @ commotion Normal Ready now open to new roles
From CI runners to service mesh to H200 GPUs — I own production platforms end to end: build, deliver, observe, and recover.
An NVIDIA H200 NVL is a lot of silicon to hand to a single notebook. Most AI workloads don't need all of it.
Multi-Instance GPU carves one card into up to seven hardware-isolated instances — dedicated memory, cache and compute per slice.
Separate training and inference clusters on Kubernetes, with MIG profiles mapped to AI/ML teams so nobody's job starves another's.
Production Kubernetes on Azure, built from scratch with Terraform. Cilium overlay networking, ArgoCD + Helm for delivery, and an Istio Ambient mesh on Gateway API in place of NGINX.
| NAME | NAMESPACE | STATUS | CHART |
|---|
Poke around. Type help.
Looking for a Senior DevOps / Platform / SRE role. Remote, Bengaluru, Hyderabad or Chennai.