DevOps Engineer — Azure DevOps CI/CD + Kubernetes (Oracle OKE) Deployments for Liferay Platform

We are launching Phase 1 of an Arabic-first web platform for a government-sector client in Saudi Arabia, built on Liferay DXP 2026.Q1 (Java 21, Gradle) with PostgreSQL 17 and Elasticsearch 8.19. We need a hands-on DevOps engineer to own the CI/CD pipeline and the UAT/Production deployments, working alongside our development team. CURRENT SETUP - Code: Azure DevOps Repos (client organisation); access provided. - CI/CD: Azure DevOps Pipelines — an azure-pipelines.yml exists with Validate and Build stages. - Cluster: Kubernetes on Oracle Cloud (OKE) — single Liferay node in UAT, two clustered nodes in Production. - Registry: Oracle Container Registry (OCIR) — image-based deployments (thin image: Liferay base + workspace WAR/JAR artifacts). - Containers: Elasticsearch Dockerfile exists (official image + Arabic/CJK analysis plugins); the Liferay app Dockerfile is in progress. - Manifests: Kubernetes manifests exist for Elasticsearch (StatefulSet, Service, ConfigMap, Secret templates); Liferay workload manifests to be authored. - Environments in scope: UAT and Production (DEV is local-only). No Terraform in scope. SCOPE OF WORK 1. Restructure the pipeline into four stages: Build Workspace, Build & Push Image (OCIR), Deploy UAT, Deploy Prod — with approvals/gates between stages. 2. Finalise the Liferay app Dockerfile (thin image) and the image tagging/promotion scheme. 3. Author/finalise Kubernetes manifests for the Liferay workload (Deployment/StatefulSet, Services, ConfigMaps, Secrets, ingress), including the 2-node clustered Production topology. 4. Secrets & config management across UAT/Prod; OCIR pull secrets. 5. Execute deployments: UAT first, then Production go-live (early August), with smoke tests and a rollback plan. 6. Document everything and hand over runbooks (deploy, rollback, scale). MUST-HAVE EXPERIENCE - Azure DevOps Pipelines (YAML, multi-stage, environments/approvals) - Production Kubernetes deployments — OKE strongly preferred (or equivalent managed K8s) - Container registries (OCIR or similar), image-based release workflows - Deploying Java/Tomcat-based platforms; Liferay experience is a strong plus - Secrets management and safe production rollouts (rolling updates, rollback) NICE TO HAVE - Liferay DXP cluster configuration (Cluster Link, sticky sessions) - Elasticsearch operations; PostgreSQL basics - Overlap with Saudi working hours (GMT+3); Arabic is a plus, not required ENGAGEMENT Hourly, 1-2 months, starting immediately — Phase 1 production launch is planned for early August, so we need someone who can move fast. Remote; NDA required before repository access. TO APPLY, PLEASE INCLUDE 1. An Azure DevOps multi-stage YAML pipeline you built (sanitised snippet or description). 2. Your OKE/OCIR (or equivalent) production deployment experience. 3. Any Liferay/Java platform deployment experience. 4. Your availability (hrs/week, start date) and timezone. 5. Briefly: how would you deploy a 2-node clustered Liferay to Kubernetes with zero downtime?