01About

I'm a Senior DevOps Engineer with 8 years in infrastructure. I started in helpdesk, and now I'm responsible for a multi-account cloud architecture. In enterprise, I worked on a Russian GDS at Amadeus scale: 500 hosts and 1500+ VMs, mission-critical, 24/7. In the Sirena-Travel group I also did SRE work: SLOs for the booking APIs, on-call with runbooks, MTTR down from 120 to 20 minutes. In a startup, I was the only DevOps engineer and built everything from scratch. Now I run 25+ EKS clusters in several regions, with GitOps via ArgoCD and FluxCD.

Backups come before features, and runbooks come before alerts. If I need a tool and it doesn't exist, I write it: FastAPI dashboards, Alertmanager webhooks, CLIs. The code goes into Git, the tool gets a container and a Helm chart. I also build AI workflows in n8n: answers to common technical questions, IaC generation, security checks of code and infrastructure. That's just how I work.

years in infrastructure
8
k8s clusters I manage
25
VMs moved to containers
1500+
booking API uptime
99.99%

02Stack

Kubernetes & GitOps
Kubernetes · k3s · k0s · Docker · FluxCD · Argo CD · Helm · Kustomize · Karpenter
IaC, CI/CD & Automation
Terraform · Ansible · GitLab CI/CD · GitHub Actions · Kaniko · n8n · Linux
Cloud Platforms
AWS · EKS · EC2 · VPC · IAM · ECR · ALB · Route53 · Yandex Cloud · MWS · Yandex Managed Kubernetes · Managed PostgreSQL · Managed Redis · Object Storage · CDN · Load BalancerYandex Cloud and MWS are Russian clouds, similar to AWS / GCP
Architecture & FinOps
Solution Architecture · Multi-Account AWS · FinOps · Cost Optimization · IAM · Cross-Region DR
Observability
Prometheus · Grafana · VictoriaMetrics · VictoriaLogs · OpenTelemetry · Vector · ELK · Loki · Sentry · Zabbix
Security & Secrets
Vault · ESO · cert-manager · SealedSecrets · Trivy · SonarQube · Keycloak · Sonatype Nexus
Networking & Ingress
Nginx · HAProxy · Traefik · APISIX · Cilium · Linkerd · MetalLB · WireGuard
Data & Messaging
PostgreSQL · MySQL · ClickHouse · StarRocks · YDB · MongoDB · Cassandra · Redis · Kafka · Strimzi · RabbitMQ · MinIO · S3
Development
Python · FastAPI · Bash · .NET · PHP · JavaScript · Go

03Experience

Jul 2026 - Present

Finstar Financial Group

Senior Cloud InfraDevOps Engineer

A fintech lending group working in Vietnam, the Philippines, South Africa, India, Sri Lanka and Europe. Each product (risk engine, communication platform, API gateway, DWH, KYC/scoring) runs on its own AWS EKS cluster. I am responsible for the reference architecture: 25+ clusters in several regions. Everything is 100% Terraform plus the shared Helm library chart, with GitOps via ArgoCD and FluxCD. Nobody does manual work in the console.

  • I am responsible for the AWS/EKS reference architecture: 25+ clusters in several regions, one per product or environment. All of them are 100% created with Terraform and deployed with GitOps (ArgoCD and FluxCD), with zero manual work in the console
  • I set up a new production EKS cluster from scratch for the data warehouse team in Africa: networking, cross-account IAM, ArgoCD bootstrap. Then I added StarRocks with its load balancer, DNS and security groups
  • I rolled out the Linkerd service mesh to new namespaces with Terraform in two regions, Africa and Europe
  • FinOps: I set up Karpenter consolidation policies and right-sized the node classes. Compute costs stay flat, even though we now have more than 25 clusters
  • I added graceful shutdown to the shared Helm library chart (terminationGracePeriodSeconds and lifecycle hooks). It now ships to 10+ clusters. I documented it in the platform guides on autoscaling, RBAC, secrets and jobs that every dev team works from
  • I built and maintain an internal Kubernetes dashboard (FastAPI + vanilla JS): workloads, RBAC, real disk usage, ingress health checks. Dev teams use it every day instead of kubectl

Kubernetes · AWS EKS · FinOps · Solution Architecture · Terraform · ArgoCD · FluxCD · Karpenter · Linkerd · Vault · External Secrets Operator · Keycloak · Kafka · Strimzi · ClickHouse · StarRocks · YDB · PostgreSQL · GitLab CI/CD · Kaniko · SonarQube · Sonatype Nexus · VictoriaMetrics · OpenTelemetry

Aug 2022 - Present · 4+ yrs

Sirena-Travel group

InfraDevOps Engineer

Four years at Sirena-Travel, across three products. On each one I ran the whole infrastructure: cloud, Kubernetes, CI/CD, observability and internal tools.

Red Rose Traveltech EU division 2026

A B2B corporate travel platform for the European market. I built the whole cloud from scratch.

  • All Yandex Cloud infrastructure is in Terraform: VPC, managed Postgres/Redis, S3, DNS, CDN, load balancers. 100% IaC, no clicks in the console
  • I set up GitLab CI/CD for each service: test → build → Helm deploy → promotion gates with auto-rollback. A commit gets to production in under 5 minutes
  • I connected Prometheus and Grafana to SLO targets, with alerts via Alertmanager and TLS from cert-manager. The booking APIs have 99.99% uptime
  • I automated recurring ops tasks with n8n: partner onboarding, SLA alerts, Terraform drift checks. Nobody has to do them by hand

MixVel GDS for travel agents 2025 - now

A real-time GDS for travel agents, with its own GDS code. I built the GitOps platform from scratch. Now CTO for the DevOps direction: I am solely responsible for the whole architecture and infrastructure.

  • I set up FluxCD v2 in hub-and-spoke mode: 6 Kubernetes clusters, where one hub cluster reconciles five spokes, with secrets in SealedSecrets. Nobody runs kubectl apply in production
  • I built a three-layer Kustomize setup (base / overlays / per-env): 90%+ of the config is shared, and there is no drift. k3s + Cilium on-prem, Yandex Managed Kubernetes, APISIX
  • I deployed the whole data layer as HelmReleases: Kafka, RabbitMQ, Cassandra, ClickHouse, MongoDB, Redis, MinIO. A new cluster takes 30 minutes with Ansible
  • Monitoring and security: kube-prometheus-stack, Victoria Logs, ELK, OpenTelemetry, Vault, Trivy Operator. I cut compute by 20% by right-sizing pods
  • I wrote the tools the team uses every day: env-view, abot, trivy-to-sonarqube, confluence-publisher. All of them are packaged as Helm charts

More in the weblog

Sirena-Travel parent GDS 2022-2026

A Russian GDS at the scale of Amadeus/Sabre: 500 hosts and 1500+ VMs, many languages (Java, Python, PHP, .NET, C++), 24/7.

  • I moved services from bare VMs to containers: XEN/libvirt → Docker Swarm → Kubernetes, 1500+ VMs on 500 hosts. Utilisation went up by 40%, with no downtime
  • I built CI/CD from scratch for every stack. Deploys used to take hours, now they take under 10 minutes
  • Promotion gates and auto-rollback cut incidents from bad deploys by 65%. I set up on-call with runbooks, and MTTR went down from 120 to 20 minutes
  • Prometheus on 500 hosts, 30+ dashboards, config in Ansible with zero drift. Quarterly OS patching with zero downtime

Kubernetes · FluxCD v2 · Kustomize · Terraform · Ansible · GitLab CI/CD · Cilium · APISIX · Kafka · Vault · Prometheus · Grafana · Yandex Cloud

Feb 2020 - Aug 2022 · 2 yrs 7 mos

ADV/web-engineering

Middle Sysadmin / Junior DevOps Engineer

I ran hybrid infrastructure for web apps on Proxmox and VMware ESXi. I brought in the first containers (Docker) and the first Kubernetes with Helm. I set up CI/CD with GitLab CI and Jenkins, managed config with Ansible, and moved monitoring from Zabbix to Grafana/Prometheus. I automated routine tasks in Bash/Python and kept the hardware running (MikroTik/HP/DELL/SuperMicro), including remote hands work.

Proxmox · VMware ESXi · Docker · Kubernetes · Ansible · GitLab CI · Jenkins · Zabbix · Grafana

Nov 2018 - Feb 2020 · 1 yr 4 mos

TenderTech

Junior / Middle Sysadmin

This is where it all started: system administration and user support (Windows/Linux, MS Office, 1C), accounts in AD/Atlassian/GitLab. I ran Linux VMs on Proxmox (Nginx, PHP-FPM) and looked after the office IT and server hardware: SuperMicro/DELL/MikroTik, AXIS/UniFi.

Proxmox · Linux · Nginx · PHP-FPM · Active Directory · MikroTik · SuperMicro

04Projects & contract work

2026

MTS contract

DevOps Engineer

MTS crypto/fintech division: blockchain settlement (VED) and two B2C exchanges. I deployed 8 environments on plain kubeadm, with GitOps via ArgoCD. I also wrote a shared GitLab CI library: Kaniko, Trivy with an OPA gate that only lets non-root containers through, SonarQube, Semgrep. Ansible handled the full cluster lifecycle, Kafka ran on Strimzi, secrets lived in Vault, and node access went over WireGuard.

Kubernetes · kubeadm · ArgoCD · Terraform · Ansible · GitLab CI · Strimzi · Vault

2024-2025

Flowerave contract

DevOps Engineer

A ticketing startup where I was the only infrastructure engineer. I built everything starting from an empty cloud account: full IaC on Yandex Cloud, k8s, self-hosted GitLab. Postgres backups with WAL archiving (RPO < 1 hr, RTO < 30 min). A new environment used to take 3 days, now it takes 30 minutes. 15 months, zero data loss.

Terraform · Kubernetes · GitLab CI · Yandex Cloud · Sentry · ELK · PostgreSQL

2023 - Present

antonnovikov.com

Personal site & homelab

The site and homelab run on one VPS (k0s, Prague). FastAPI + Jinja2 behind APISIX, per-host TLS from Let's Encrypt via cert-manager, a private Docker registry, WireGuard and IKEv2, HTTP/SOCKS5/Shadowsocks proxies with health checks. A deploy is one command (build → push → rollout). Full observability (metrics, logs, alerts, dashboards) fits into ~452 MiB of memory requests. A dozen more services run on the same VPS, and the whole thing costs ~€130 a year.

More in the weblog

k0s · FastAPI · APISIX · cert-manager · WireGuard · Litestream

2026 - Present

Weblog & Cheatsheet

Technical write-ups from real infrastructure: FluxCD, observability, security, self-hosting. Not retold docs, but solutions that made it to production. Plus a working cheatsheet with 1600+ commands for 100+ tools (kubectl, helm, terraform, vault, linux and more), which I keep up to date.

Kubernetes · Observability · Security · Self-hosting