πŸš€ DevOps / Cloud / AI Infra 🧠 Team Lead @ Gcore Cloud πŸ“ Nicosia, Cyprus

Aleks

Infra that doesn't cry: Kubernetes, RDMA/InfiniBand, H100, Crossplane, Grafana, vLLM, GPU-aware scheduling. In love with simple architectures and strict SLOs.

Email me β†— GitHub LinkedIn

βš™οΈ What I do

I build and tame cloud infrastructure. K8s, networking, storage, observability, pipelines. I make it fast and painless for engineers.

🧩 What I tinker with

AI infra: training/inference, LoRA/SFT, vector DBs, vLLM. I love optimizing the GPU stack, NUMA, RDMA.

πŸ“ˆ KPI, but humane

99.95% SLO TTR ↓ x2 Cost ↓ 20–40% Deploys ↑ daily
Infra without tears

Clean interfaces. Predictable releases. Sensible alerts. Documentation people actually read.

πŸ›  Technologies

Kubernetes, Cilium, Helm, Crossplane, Terraform, ArgoCD, Prometheus/Grafana/VictoriaMetrics, Nix, GitHub/GitLab CI, KubeVirt, SR-IOV, IB/RDMA, NV GPU (H100), MLflow, vLLM.

πŸ’¬ Talks / threads

About Kubernetes networking, admission policies, observability, AI infra. Links later β†’

🚧 Projects / Threads

πŸ”₯ AptenBot

Telegram proxy-bot for LLM providers. Minimal, stylish, vibey.

Try

πŸ“Š Preheat Controller

TODO: K8s controller to pre-warm images. Less pain during deploys.

GitHub

🧡 Threads

A collection of posts about K8s, Observability, and AI infra.

Twitter/X