Daniil Z.

Senior DevOps / SRE engineer and cloud architect with 20 years in infrastructure. I design, build, migrate and run production platforms on AWS and Google Cloud: Kubernetes (EKS, GKE, Rancher), Terraform / OpenTofu, GitOps with ArgoCD, and observability that catches incidents before your customers do. Top Rated Plus · 100% Job Success · $300K+ earned · 44 jobs · 6,100+ hours on Upwork SELECTED WORK (real numbers; client names withheld under NDA) ▸ Cloud cost: cut a client's GCP bill by about a quarter (~€3,400/month) through workload right-sizing, storage lifecycle policies and egress optimisation. Every saving was measured and documented before anything was touched. ▸ Production migration: moved a real-time communications platform (23 containers, PostgreSQL + ClickHouse, 8 worker replicas) between hosting providers. Staged cutover in a planned low-traffic window, old server kept as a live rollback path for 7 days, monitoring parity verified before and after. Next day: 98.9% job success on ~31k jobs/day. ▸ Disaster recovery: zero-standby-cost DR environment in OpenTofu that provisions in under a minute and tears down in 30 seconds, drilled end to end for under €1 per drill. The first drill caught the restore path reading a stale backup; fixed and written into the runbook the same day. ▸ Kubernetes platform: lead DevOps for a video distribution platform. Five Rancher-managed clusters, GitLab CI pipelines, HashiCorp Vault for secrets, all infrastructure in OpenTofu. ▸ GCP long-term: built and ran GKE + Cloud SQL production for a SaaS client for over two years ("the results have been exceptional", 5-star, 150+ hours). ▸ SOC 2: took a SaaS company's AWS estate from 15% to 100% in Drata and through a passed audit, all in Terraform (VPC, IAM, KMS, CloudTrail / GuardDuty, EKS, RDS), then helped hire and hand over to their in-house DevOps. ▸ AWS takeover: inherited a cashless payments platform (ECS Fargate, RDS, Terraform) after the CTO left. Closed a bastion open to the whole internet, put 15 CloudWatch alarms in place, and built an off-season shutdown / restore playbook that cut the bill from ~$835 to ~$154/month while idle. WHAT I DO ▸ Cloud architecture & migrations: AWS, GCP, Azure, Hetzner. Landing zones, networking, IAM, high availability and DR; moves between providers and from bare metal into Kubernetes. ▸ Kubernetes platform engineering: EKS, GKE, K3s, RKE2 / Rancher, Helm, autoscaling (HPA, KEDA), ingress, network policies, GitOps with ArgoCD. ▸ Infrastructure as Code: Terraform / OpenTofu, Pulumi, Ansible, CloudFormation. Plan reviewed in CI before apply, nothing changed by hand. ▸ CI/CD: GitLab CI, GitHub Actions, Jenkins, Bitbucket Pipelines; rolling, canary and blue/green releases. ▸ SRE & observability: Prometheus, Grafana, Loki, Datadog, CloudWatch, Sentry. Alerting wired to Slack / PagerDuty that only pages on what matters; incident response and runbooks. ▸ Security & compliance: HashiCorp Vault, SOPS / age, least-privilege IAM, VPC design, Tailscale, SOC 2 readiness with Drata. ▸ Databases in production: PostgreSQL, MySQL, Redis, MongoDB, ClickHouse. Backups, replicas, restore drills, zero-downtime schema migrations. ▸ FinOps: right-sizing, committed-use discounts, lifecycle policies, per-team cost visibility. Measured, not guessed. ▸ AI in operations: LLM-assisted runbooks and agentic workflows on Temporal, self-hosted model serving on your own GPUs. AI in your ops without sending your data to third parties. Application side: I also write Python / FastAPI and Node.js / TypeScript, so I can fix the application half of a deployment problem instead of handing it back to your developers. HOW I WORK ▸ Rehearse, then execute: drills, dry runs and a written rollback plan before anything touches production. ▸ Measure first: cost and performance changes are found, documented and agreed before I change anything. ▸ Your infrastructure stays yours: everything lives in your accounts and repos, as code, with handover docs. Nobody should need me to keep the lights on. ▸ Honest scoping: you get the risks and a realistic timeline before hours are billed. If something is out of my depth, I say so. ▸ Clear written updates in your channel. Based in Barcelona (CET), overlapping EU and US East Coast hours. Fluent English and Spanish, native Russian. A GOOD FIT IF YOU ARE • A startup that needs a production-grade cloud foundation and CI/CD in weeks, not months • A scale-up fighting Kubernetes reliability, a growing cloud bill, or a SOC 2 deadline • A team that lost its DevOps or CTO and needs someone to take over infrastructure safely • A company that needs a senior DevOps / SRE or cloud architect without a full-time hire If your deploys are slow, your cloud bill keeps growing, or you're one incident away from finding out you have no monitoring, tell me what's going on and I'll come back with a concrete plan.