Building reliable

GPU infrastructure.

Hi, I'm Ezgi, an AI Infrastructure Engineer specializing in GPU computing, Kubernetes, and cloud-native ML systems.

[email protected]:~ 
$ whoami
Ezgi Tastan
# AI Infrastructure Engineer

$ cat ~/.config/interests.toml
gpu_infra     = "MIG Slicing, GPU Operator, DCGM, NCCL"
orchestration = "Kubernetes, OpenShift, Helm"
ai_ml         = "vLLM, KServe, LLM Inference"
cloud         = "AWS, GCP, Terraform, Terragrunt"
observability = "Datadog, Prometheus"
security      = "SOC2, Trivy, SecComp, RBAC"
Presentations ::~/gpu

Latest Posts

My writings about GPU infrastructure, Kubernetes, and AI systems.

Observability

Profiling MiniMax-H3 on GB300 with SGLang and eBPF

I used eBPF to profile MiniMax-H3 on SGLang and a GB300 tray, measured stage timing, Ulysses, concurrency, fabric traffic and energy.

Profiling MiniMax-H3 on GB300 with SGLang and eBPF
Observability

Profiling Voice AI on H100 with eBPF

I traced STT and TTS containers on an H100 with bpftime. The GPU was at 0% while the CPU was doing 169K malloc/s. Here's everything I found.