Home · Experience · Projects · Services · Contact · Resume (PDF)
Yousaf Hamza: Senior DevOps Engineer in Dubai, UAE
Yousaf Hamza (also Yousaf K Hamza) is a Senior DevOps / DevSecOps & Platform Engineer based in Dubai, UAE with 8+ years of experience. Senior DevOps & Platform Engineer in Dubai. Runs secure, highly available AWS and Kubernetes platforms for fintech and regulated enterprises: EKS migrations, Terraform automation, GitOps delivery, and SOC 2 / PCI DSS / ISO 27001 controls built into CI/CD.
Experience
- Senior DevOps Engineer, Pay10 UAE (Oct 2025 – Present)
- Senior DevOps Engineer, Alamy (Nov 2023 – Oct 2025)
- Senior Platform Engineer (SRE), Quantiphi (Aug 2021 – Nov 2023)
- Site Reliability Engineer, Armia Systems (Jun 2020 – Aug 2021)
- TechOps Engineer, Activelobby IS (Oct 2018 – Jun 2020)
Skills
- Cloud: AWS, EKS, EC2, IAM, RDS, S3, CloudTrail, AWS Organizations, Azure (AKS, Entra ID), GCP (GKE)
- Containers: Kubernetes, Docker, Helm, Karpenter, HPA / VPA, Amazon ECS, AKS, GKE, OpenShift
- IaC & scripting: Terraform, Bash, Python, CloudFormation, Ansible, Go (CLI tools), TypeScript
- CI/CD & GitOps: Argo CD, GitHub Actions, Jenkins, Kustomize, GitLab CI, Bitbucket Pipelines, Concourse CI, k6 load tests
- Security & compliance: GuardDuty, Wazuh, Trivy, JumpServer PAM, HashiCorp Vault, Checkov, SonarQube, PCI DSS, SOC 2, ISO 27001
- Observability: Prometheus, Grafana, Loki, Grafana Alloy, CloudWatch, OpenTelemetry, Fluent Bit, ELK, New Relic
- AI & LLM Ops: Agentic AI, RAG, Ollama, Qwen, Claude API, Amazon Bedrock, Qdrant, vLLM, GPU node pools, OpenAI / Gemini APIs
- Networking & access: VPC, Route 53, AWS WAF, Site-to-Site VPN, OpenVPN, Direct Connect, NAT Gateway, Squid proxy, SSO (Entra ID)
- Data & streaming: PostgreSQL, MySQL, MSSQL, MongoDB, DynamoDB, Apache Kafka, Kinesis
Open-source projects
- K8s Incident Copilot: Know why a Kubernetes pod is failing before you open a terminal. Runs in your cluster, and when a pod breaks it gathers the spec, events, logs from the crashed run and volume claims, then has a model explain the root cause and the fix and sends it to Slack, Teams, email or any webhook. Uses a local Ollama model by default so nothing leaves the cluster, or Claude, OpenAI, Gemini and Bedrock as an agent with read-only cluster tools. It never reads Secrets, can never change a workload, and redacts tokens and keys before anything is sent.
- platform-copilot: A self-hosted RAG assistant that answers questions from your own documents, with Qdrant for vector search, Ollama for local inference and FastAPI tying it together. Every layer runs inside your own cluster, so embeddings, vectors and the LLM never leave it. Chunks documents at ~800 characters with overlap so a fact split across a boundary isn't lost, and forces the model to answer only from retrieved context, citing the source file so an answer can be checked instead of trusted blindly.
- DevOps Quiz 2.0: 678 DevOps questions across 16 topics, from quick concepts to multi-step production incidents you fix by typing real commands. Questions come as real scenarios with logs, kubectl output and manifests, in six formats: MCQ, multi-select, live terminal labs, manifest fill-ins, spot the bug and order the runbook. Exam mode adds timers and negative marking, practice mode gives instant feedback, and results break down by topic and level. Fully static on GitHub Pages, so nothing leaves the browser.
- IAM Policy Visualizer: A client-side tool that parses AWS IAM policy JSON and renders each statement as a risk-scored card. Flags wildcard actions, unscoped resources, missing Condition blocks and privilege-escalation paths like unconditioned iam:PassRole, all explained in plain English. Built for reviewing a policy before it ships rather than after an incident. No policy data ever leaves the browser.
- iso27001-auditor: A read-only AWS scanner that checks EC2, RDS, EKS, ECS, IAM, S3 and a dozen other services against ISO 27001:2022 Annex A controls. Produces an evidence report rather than a pass/fail badge: every finding carries the exact resource ID and a remediation command, so it doubles as audit evidence and a to-do list. Runs from a single read-only role, and a missing permission or an unused service is logged and skipped instead of crashing the scan or showing a false failure.
- podgauge: A Kubernetes CPU/memory right-sizing tool that reads like a server-rack status wall instead of a spreadsheet. A read-only script diffs kubectl top usage against each Deployment's requests and limits, then you paste the CSV into the browser and nothing leaves the tab. Suggests right-sized requests and a memory limit with real headroom, but deliberately skips a formula-based CPU limit by default, since one sized off steady-state usage throttles JVM startup bursts and turns a 10-20s boot into minutes.
- Universal App Helm Chart: A production-ready, reusable Helm chart for deploying applications on Kubernetes and AWS EKS. Supports Deployments, StatefulSets, ALB Ingress, HPA, EBS and EFS volumes, ConfigMaps, Secrets, IRSA, ServiceAccounts, node scheduling and health probes. Everything is configured through a single values.yaml, and it ships with production-ready examples so a new service can go from zero to deployed without writing a chart of its own.
- kubesizer: A client-side tool that parses kubectl top output and renders sized CPU/memory requests and limits per container. Uses peak plus buffer for a single snapshot, and switches to percentile-based sizing (p50 requests, p95 limits) automatically once multiple snapshots are pasted. Ships with a capture script for gathering samples during peak load, and shows exactly which snapshot a container's data is missing from if a kubectl call fails mid-run. No cluster data ever leaves the browser.
- loadcannon: A CLI that load-tests internal and public HTTP APIs from one scenario format: public hostname, internal load balancer, or a direct IP with a Host-header override. Wraps k6 for the traffic itself while loadcannon handles target resolution and secret injection. Tokens are resolved at runtime from an env var, AWS SSM or a masked prompt, and only ever passed to k6 through its process environment, never as an argument or on disk. Six live example scenarios run end to end with zero setup.
- aws patch: A production-grade Linux patch automation tool for AWS EC2 fleets, built in Bash. Supports Ubuntu, Amazon Linux 2 and Amazon Linux 2023 across multiple regions, handling apt, yum and dnf from a single command. The codebase is ShellCheck-clean with 51 passing tests, and a one-line installer gets it running on a new machine in seconds, so patching a fleet stops being a manual, host-by-host chore.
- accessrig: A one-command CLI wrapper around JumpServer's official installer, built after one too many manual PAM deployments went sideways. Detects install versus update from real container state rather than a bookkeeping file that can drift, and pins an explicit version on every run instead of silently pulling latest. Downgrades need an explicit flag on top of an automatic backup, and DOMAINS is set at install time to avoid the common 'CSRF Failed: Origin checking failed' error behind a real domain.
- openvpn-dns-tools: A DNS-attribution toolkit for OpenVPN that answers what OpenVPN itself can't: who ran this query. Captures DNS on the tunnel interface and correlates it with live session state from Community Edition or Access Server, auto-detecting which one you run. VPN IPs get reused, so every match is resolved by IP and timestamp together, never IP alone. A one-command installer detects the tunnel interface and backend, then wires up short commands and systemd services.
- The Manifest: A self-publishing daily e-newspaper for cloud and infrastructure news covering AWS, Azure, GCP, Kubernetes, Terraform, Python, Go, Rust and Node.js. A GitHub Actions cron job pulls each topic's official RSS/Atom feed every morning and keeps only the publisher's own title, summary and thumbnail, falling back to original cover art when a feed has none. The broadsheet-style front page rebuilds itself every day with zero manual steps.
- pxtrim: A tiny, fully client-side tool to resize, batch convert and compress images. Drop one file or a whole batch, adjust scale, quality and format, and download the results. There are no uploads, no backend and no storage: everything runs in the browser through the Canvas API and only for the current session, so images never leave your device.
Certifications
- Certified Kubernetes Administrator (The Linux Foundation, 2024)
- AWS Certified Solutions Architect – Associate (Amazon Web Services, 2023)
- HashiCorp Certified: Terraform Associate (HashiCorp, 2022)
- Generative AI Foundations (with Microsoft) (upGrad, 2025)
- LFS169: Introduction to GitOps (The Linux Foundation, 2024)
- Fundamentals of Ethical Hacking (Offense Hackers Academy, 2023)
- OCI Foundations Associate (Oracle, 2022)
- Python (Basic) Skills Certificate (HackerRank, 2021)
- Verified International Academic Qualifications (World Education Services, 2022)
Email: yousaf.k.hamza@gmail.com · LinkedIn · GitHub