Kubernetes & Platform
I run Kubernetes clusters (kubeadm, Kubespray, Talos Linux) up to 30 nodes. Migrated workloads from Docker Swarm to Kubernetes following a GitOps model, set up dynamic staging via Helm + CI/CD.
about
Aleksandr
Suprun
I'm a DevOps / Platform Engineer with 5+ years across bare-metal, AWS/GCP, and hybrid infrastructure. I went from Linux operations on a 1200+ server fleet across three data centers, through cloud optimization and IaC automation, to Kubernetes platform engineering, ML/GPU infrastructure, and SRE practice. AWS and HashiCorp certified.
Core domains
ML/GPU & CI/CD
I run GPU-enabled Kubernetes nodes with MIG and time-slicing for ML/LLM workloads. Designed GitLab CI/CD pipeline architecture with templating and prebuilt base images. Cut build times by 40% with a caching strategy.
SRE & Observability
I own disaster recovery with explicit RTO/RPO. Automated backup and validation for PostgreSQL, MongoDB, ClickHouse, Elasticsearch. Built SRE dashboards with SLI/SLO on the VictoriaMetrics stack with Grafana, Zabbix, AlertManager.
Security & HA
I designed a custom KMS server for LUKS disk encryption with heartbeat-based node validation. Rolled out centralized SSO via Keycloak across GitLab, Grafana, Vault. Built multi-DC HA with HAProxy and virtual-IP failover.
My Notes section is a personal interview-prep and working-reference archive — not a tutorial library and not a substitute for vendor documentation.
This is my portfolio and personal knowledge base — built on Next.js, automated with AI agents.