SC
Open to opportunities

Sylvain Costes

Platform Engineer · Azure / AKS · Terraform · CI/CD

About

I'm a Platform Engineer at Société Générale CIB, where I build and maintain the infrastructure that powers critical payment platforms (SEPA, Cross-Border). Every day I work on systems where reliability is non-negotiable: a pipeline failing in prod means real financial transactions on the line.

What drives me? Turning deployment friction into something invisible for product teams. I firmly believe DevOps culture is first and foremost about empathy for developers: understanding their constraints before pushing tools at them.

Before specialising in Platform Engineering, I worked as a Full-Stack Engineer (Java / React / AWS Serverless) and shipped a production LLM chatbot at Optic 2000. That dual background — application code + infrastructure — lets me design CI/CD pipelines that reflect the realities of the field.

Outside of work, I run a homelab on Raspberry Pi (Docker, Portainer, Traefik reverse proxy, self-hosted CI/CD) — my sandbox for testing without a safety net what I apply at work. Hackathons, CTFs, conferences (AWS Summit Paris): I stay in a permanent learning mode.

Master MIAGE (Paris Nanterre, with distinction) · AWS Certified Cloud Practitioner · TOEIC 935/990

Main Stack

Kubernetes / AKS
Terraform
ArgoCD
Helm
Docker
Azure
AWS
Bash / Shell
Python
GitHub Actions

Timeline

  1. Platform / DevOps Engineer

    Société Générale CIB

    Critical payment platforms (SEPA / Cross-Border). GitOps CI/CD pipelines serving ~40 engineers across France and India. Stack: AKS, ArgoCD, Helm, GitHub Actions, Terraform.

  2. Full-Stack Engineer (apprenticeship)

    Optic 2000 / ATS

    Built and shipped a production LLM chatbot (Amazon Bedrock / Claude). Serverless architecture on AWS (Lambda, S3). Automated testing with Cypress.

  3. Co-founder & DevOps

    GreenTrip

    CI/CD infrastructure, GitHub Actions, end-to-end deployment pipeline setup for the project.

Reliability first

A system that goes down in prod costs more than a delayed feature. I optimise for availability and observability before velocity.

Everything is code

Infrastructure, configuration, runbooks: if it's not in a repo with a PR, it doesn't exist. GitOps isn't a buzzword, it's hygiene.

Permanent curiosity

CTFs, hackathons, homelab: I deliberately break things to understand how they work. That's what separates someone who follows tutorials from someone who resolves incidents at 3am.