GitOps everywhere, Kubernetes at scale, and a stubborn refusal to click buttons twice.
I keep large, boring, extremely reliable systems boring. Day to day that means Kubernetes fleets, Terraform/Terragrunt modules, Prometheus/Thanos, and enough Go and Bash to remove the toil before it becomes a runbook.
role: Senior SRE / Platform Engineer @ Equativ
based_in: France 🇫🇷
focus: [ kubernetes, gitops, observability, finops, automation ]
scale: { servers: ~4500, clusters: ~30, qps: 2.5M, regions: 5 }
currently: writing kubectl plugins and Prometheus exporters in Go
fuel: coffee, travel, and cooking way too much food| 🖥️ ~4 500 servers | OVH, LeaseWeb, AWS, managed as code |
| ☸️ ~30 Kubernetes clusters | on-prem (Rancher, RKE2) + GKE / EKS |
| ⚡ 2.5M QPS | worldwide adserving traffic, low latency |
| 🌍 5 regions | EU, NA, APAC |
| 📈 Thanos | long-term metrics across every cluster |
|
Read-only cluster inspection as a kubectl plugin. One dispatcher, ~25 shortcuts, shipped through my own krew index.
|
🏗️ kubearchPrometheus exporter reporting the CPU architectures every running image supports. Reads OCI manifests only, never pulls a layer. Built for arm64 migrations.
|
🌐 gopenOpen the current git repo in the browser at the right branch, file and line. GitHub, GitLab, Bitbucket, Azure DevOps, Gitea. Zero dependencies.
|
My public knowledge base: Kubernetes, Terraform, observability, incident notes. MkDocs Material, deployed on every push.
|
|
Server-authoritative online UNO for 2 to 4 players. Cheating is structurally impossible: opponents' cards never cross the wire. Play it.
|
Scores and ranks unread FreshRSS articles by topic relevance (SRE, K8s, Terraform). FastAPI + SQLite, packaged with Helm.
|
🧰 More things I maintain
| Project | What it does | Lang |
|---|---|---|
| dotfiles | Full macOS setup, automated from scratch | Shell |
| prometheus-rules | Alerting rules for common exporters | YAML |
| nginx-ensite | a2ensite / a2dissite equivalents for nginx |
Shell |
| krew-index | My custom krew index for kubectl plugins | - |
| homebrew-tap | Homebrew tap for my CLI tools | Ruby |
| tcp_rtt_exporter | TCP round-trip time as Prometheus metrics | Python |
| dockerfiles | Container images I actually run | Dockerfile |
Cloud & Infra
Kubernetes & GitOps
Observability
Code & Data
Site Reliability Engineer · Equativ · Oct 2022 → present
- Run ~4 500 servers across OVH, LeaseWeb and AWS in 5 regions, entirely through GitOps
- Operate ~30 on-prem Kubernetes clusters (Rancher, RKE2, ArgoCD) plus GKE and EKS
- Keep 2.5M QPS of adserving traffic fast and available, with Prometheus/Thanos and Grafana OnCall
- Own GCP infrastructure as code: GKE, CloudSQL, GCS, Cloud DNS, via Terraform and Terragrunt
- Built secret management on HashiCorp Vault for credentials and sensitive config
FinOps & Azure
- Cloud spend analysis and rightsizing, budgets enforced with engineering and finance
- Provisioning and cost tracking automated with Terraform and Ansible
🌏 Travel for the places I have not broken yet · 🍳 Cooking, with the recipes published at cuisine.jdelgado.fr · 🎮 building silly multiplayer games on weekends
Reproducible, observable, and cheap. Pick three.