11+
Years in platform & SRE
Cloud-native infrastructure and team leadership
12 20 October 2026
Shankar Ramanathan · Portfolio
Platform
as a product.
Platform Leader SRE Architect CKA Open to opportunities · GMT+5:30
§ I. Field notes + + +
Eleven+ years scaling cloud-native infrastructure and building platform teams. Recent leadership work has been architecting self-service Kubernetes environments for large SaaS platforms: 100+ microservices, 100 Kubernetes clusters, and 3k+ compute instances. I treat the internal platform as a product paved roads, policy-as-code, and roadmaps that line up with reliability, security, and FinOps.
Hands-on through GitOps with Crossplane and Argo CD, progressive delivery with KubeVela and Argo Rollouts, Karpenter/KEDA autoscaling, Kyverno guardrails, and deep observability. Open to Platform Engineering/SRE Leadership roles. Based in GMT+5:30.
100
Kubernetes clusters
GitOps provisioning, canary-to-GA waves
100+
Microservices on the platform
Self-service Kubernetes for SaaS scale
3k+
Compute instances
Spot, Graviton, and right-sized node pools
80%
Faster CI/CD cycles
KubeVela · Argo Rollouts · ApplicationSets
99.8%
Change success rate
Progressive delivery with tight blast radius
§ II. Blogs + + +
№ 01
ARC zonal shift on Amazon EKS
ARC zonal shift moves traffic off an impaired Availability Zone in minutes — but EKS data plane, Karpenter, and capacity planning still have to keep up. Preparatory steps, failure modes, and two capacity patterns that work under a shift.
→
№ 02
Stop Renting GPUs You Do Not Fill Part1 - The Math Behind Amazon Bedrock, Frontier Models & Hybrid Model Routing
Managed APIs vs. Self-Hosted Silicon, Tipping Point - When to Move Beyond Amazon Bedrock to Hybrid Inference
→
№ 03
Stop Renting GPUs You Do Not Fill Part2 - Enterprise Governance with RBAC, Kyverno & Bedrock Inference Profiles
Governing gen-AI platform - Routing Multi-Provider Inference with LiteLLM, Kyverno, Crossplane on EKS
→
№ 04
Kubernetes(EKS) control plane scales gracefully, Until you hit Scaling issues
Kubernetes(EKS) Control Plane scaling walls that you must know when running Clusters @ Scale
→
№ 05
Argo Rollouts + AnalysisTemplates ~ Release with Confidence
Replace a Kubernetes Deployment with an Argo Rollout CRD, shift traffic in steps, and let Analysis steps decide promote vs abort — metric-gated canaries instead of hope-based rollouts.
→
№ 06
Kubevela — the abstraction Dev teams need
Reliable & re-usable application delivery via the Open Application Model. KubeVela gives platform guardrails without forcing every team to hand-write Ingress, Services, and Autoscalers.
→
§ III. Accomplishment + + +
EKS & GitOps at scale
Crossplane + Argo CD provisioning, 90-minute cluster setup, canary-to-GA waves that cap blast radius at 10–15 clusters.
FinOps & autoscaling
Karpenter and KEDA for 2–3x surges, scheduling in tens of seconds, ~30% per-core efficiency via Spot and Graviton.
Progressive delivery
80% faster CI/CD with KubeVela, Argo Rollouts, and ApplicationSets. 99.8% change success rate.
Security & supply chain
Kyverno policy-as-code guardrails. Chainguard base images with SBOM support and SLSA provenance.
Reliability & observability
P1 median MTTR under 60 minutes, 35% fewer repeat critical incidents. Prometheus, Grafana, VictoriaMetrics.
Team scaling & mentorship
Structured hiring pipelines and onboarding frameworks so new platform engineers reach productivity faster.