00
home
Aleksandr
Suprun
Senior DevOps / Platform Engineer — bare-metal Kubernetes at scale, GPU infra for ML/LLM, multi-DC HA, custom security tooling.
01
I design and operate Kubernetes platforms (kubeadm, Kubespray, Talos) up to 30 nodes with GPU MIG and time-slicing for ML/LLM workloads. Background: a 1200-server multi-DC fleet, on-call rotations included.
02
I own the SRE side: SLI/SLO, multi-DC HA via HAProxy + VIP failover; observability on the VictoriaMetrics stack with Grafana. Designed a custom KMS for LUKS disk encryption and rolled out centralized SSO via Keycloak + Vault.
03
Daily-driving Debian with i3wm — minimalism, control, reproducibility.