Client : ENGIEPlatform & CloudPeriod : Jul 2024 — present
Ephemeral GitHub Actions runners (ARC), in-cluster Jenkins controllers, and in-house tooling for a DevOps platform serving 600+ organizations and 10,000+ users.
600+
organizations served by the platform
10 000+
daily users
−60 %
intervention time thanks to automation
ENGIE Digital & IT operates the group's DevOps platform — GitHub Enterprise, Jenkins, Artifactory/Xray, SonarQube, Jira/Confluence — for 600+ organizations and 10,000+ users. On the infrastructure side: 6 production Kubernetes (EKS) clusters, 200+ VMs, and 97 Terraform stacks spread across multiple AWS accounts.
At this scale, every manual action costs a hundredfold. The challenge: provide reliable, elastic build capacity to hundreds of organizations, keep node sizing under control, and cut the time swallowed by recurring interventions.
Ephemeral GitHub Actions runners deployed at scale with Actions Runner Controller (ARC), directly in the EKS clusters.
In-cluster Jenkins controllers for the legacy CI pipelines.
Migration from Cluster Autoscaler to Karpenter for node provisioning.
Custom internal tooling: alert centralization in Python, runner management in Go, a monitoring application in Next.js + Go.
Systematic automation of recurring interventions.
Work
Describe your context — reply within 24 to 48 h.
Neurones IT · Platform & Cloud
Design and technical lead of a multi-tenant platform for creating and managing Kubernetes clusters — Vanilla, K3S, RKE/RKE2, OpenShift — on AWS, GCP, Azure, and OVH.
4
clouds covered — AWS, GCP, Azure, and OVH
Go · Pulumi · Ansible · Next.js +8
View the case study
ENGIE · Platform & Cloud
A complete Grafana stack — Alloy and Vector for collection, Mimir, Loki, and Tempo for storage — deployed across multiple clusters on ENGIE's DevOps platform.
6
production EKS clusters covered
Grafana · Alloy · Vector · Prometheus +4
View the case study