DevOps Engineer Interview Questions India 2026: Complete Guide (Kubernetes, CI/CD, Cloud)

Five DevOps interview question categories: (1) Kubernetes depth ('Explain the difference between a Deployment and a StatefulSet and when you would use each', 'How do you do a zero-downtime rolling upgrade in Kubernetes?', 'What is a PodDisruptionBudget and when would you use it?', 'Explain how Kubernetes networking works: Services, Ingress, and CNI plugins'). (2) CI/CD ('Describe your ideal CI/CD pipeline from code commit to production', 'How do you handle secrets in a CI/CD pipeline securely?', 'What is GitOps and how does ArgoCD implement it?', 'How do you implement canary deployments with Argo Rollouts?'). (3) Infrastructure as code ('Write a Terraform module that provisions an ECS cluster with an Application Load Balancer', 'What is the difference between Terraform apply and plan?', 'How do you manage Terraform state in a team environment?'). (4) Cloud platforms ('Explain VPC peering vs PrivateLink in AWS', 'How does AWS IAM work? Explain roles vs policies vs permission boundaries', 'What is the difference between SQS FIFO and Standard queues?'). (5) Monitoring and observability ('Explain the three pillars of observability: metrics, logs, traces', 'How would you set up an alert for high error rate that avoids alert fatigue?', 'Explain the USE method and the RED method for monitoring services').

Kubernetes depth tested at mid and senior DevOps/SRE interviews: Kubernetes architecture (control plane: API server as the single truth, etcd as the distributed key-value store for cluster state, scheduler for assigning pods to nodes, controller manager running the reconciliation loops for Deployments, ReplicaSets, and Services; data plane: kubelet as the node agent, kube-proxy for Service IP routing via iptables or IPVS, the container runtime: containerd). Pod scheduling (NodeSelector vs NodeAffinity vs Taints and Tolerations for directing pods to specific nodes: when to use each; PodAffinity for co-locating pods that communicate frequently; anti-affinity for spreading replicas across nodes for HA). Resource management (Requests and Limits: how CPU throttling works (CPU compressible: throttled but not killed); how memory works (OOMKilled when limit exceeded); LimitRange and ResourceQuota for namespace-level governance; HPA (Horizontal Pod Autoscaler) based on CPU/memory or custom metrics via Prometheus Adapter; VPA (Vertical Pod Autoscaler) for right-sizing pod resource requests automatically). Kubernetes networking (ClusterIP for internal service discovery; NodePort for external access without a load balancer; LoadBalancer for cloud-provider-provisioned external load balancers; Ingress for HTTP/HTTPS routing with a controller like nginx-ingress or AWS ALB Ingress Controller; Network Policies for pod-level firewall rules).

CI/CD pipeline depth: GitHub Actions (the dominant CI/CD choice for new projects in India: workflow syntax (`.github/workflows/`), jobs, steps, matrix builds for testing across multiple Node/Python versions, caching (actions/cache for npm or pip dependencies), environment secrets (repository vs environment secrets), OIDC (OpenID Connect) for passwordless AWS authentication from GitHub Actions, and reusable workflows for DRY pipeline configuration). ArgoCD and GitOps (GitOps principle: Git is the single source of truth for cluster state; ArgoCD synchronises Kubernetes cluster state to match the Git repository; Application resource: defines source (Git repo + path) and destination (cluster + namespace); AppProject for RBAC and multi-tenancy; ApplicationSet for templating ArgoCD Applications across multiple clusters or environments). Progressive delivery (Argo Rollouts: canary deployment (route 10% of traffic to new version, monitor error rate, auto-promote or auto-rollback based on Prometheus metrics) and blue-green deployment (switch traffic atomically between old and new Deployment); Flagger as an alternative for Istio-integrated canary). Secret management in CI/CD (HashiCorp Vault with Vault Agent Injector or External Secrets Operator for syncing Vault secrets into Kubernetes Secrets; AWS Secrets Manager with ExternalSecrets; SOPS (Secrets OPerationS) for encrypting secrets in Git with KMS).

DevOps/SRE salary ranges: junior (0-2 years): Rs 6-18 LPA. Mid-level (2-5 years): Rs 15-45 LPA. Senior (5-8 years): Rs 40-90 LPA. Staff/Principal SRE (8+ years): Rs 80-160 LPA. Top-paying DevOps employers: Google SRE India (Rs 60-160 LPA mid-senior: the highest-paying DevOps-adjacent roles in India), Amazon India (Rs 40-120 LPA: AWS infrastructure, Prime Now DevOps), Flipkart (Rs 25-70 LPA: one of India's largest Kubernetes deployments, 10,000+ node clusters), Razorpay (Rs 25-65 LPA: payments infrastructure reliability is mission-critical), CRED (Rs 25-65 LPA). Cloud-certified engineers (AWS Solutions Architect Professional, CKA, GCP Professional Cloud Architect) earn 15-25% more than non-certified peers at equivalent levels. The SRE path at Google and Amazon typically pays 30-40% more than comparable DevOps roles at Indian product companies.