Société Générale

Context

Within a cross-functional SRE team, I worked with development and architecture teams on infrastructure delivery, automation, deployments, and application stability. The team provided SRE services and platform support for a suite of structured-finance applications deployed in Europe, the Americas, and Asia.

Standardizing backend and batch deployments with Helm

The problem

Kubernetes deployment manifests were generated by an in-house Python solution that reproduced capabilities already available in Helm. Changes required maintaining custom generation logic across the supported deployment cases, making the solution complex to evolve.

The internal tooling also introduced an onboarding burden: engineers had to learn a bespoke solution rather than apply knowledge of a tool already familiar to the development and DevOps community.

My contribution

My main contribution to the migration was designing and implementing Helm charts for backend services and batch workloads, replacing their dependency on the internal manifest-generation approach.

The outcome

Deployment updates now rely on Helm charts to encapsulate the configuration and templating complexity, moving changes into a standard chart structure rather than bespoke Python generation logic.

Engineers familiar with Helm could use their existing knowledge instead of learning the in-house solution. The charts also fit the team’s Argo CD GitOps delivery workflow.

Delivery automation and GitOps

• Automated weekly build and deployment cycles for stable releases, including SRE library updates, using Python tooling.
• Contributed to the migration from Jenkins shared-library pipelines to standardized GitHub Actions workflows.
• Reworked continuous delivery around GitOps with Argo CD across environments.

Reliability and observability

• Optimized Kubernetes resource configurations for Spring Boot microservices and monitored their performance impact.
• Centralized frontend microservice logs and monitoring with Fluent Bit.
• Upgraded the ingress controller for compatibility and maintainability.
• Migrated a managed Elasticsearch infrastructure, using scripted validation and monitoring with Terraform, Ansible, Docker, and Docker Compose.

Secrets, image security, and maintenance

• Refactored and migrated Vault KV secrets across microservices.
• Implemented secret rotation and certificate renewal through init containers.
• Created and standardized secure Docker images for frontend services.
• Developed automated Artifactory registry cleanup processes.

Kubernetes automation proof of concept

I also implemented a Kubernetes automation proof of concept on GCP.

• Provisioned infrastructure with Terraform and configured clusters with Ansible.
• Designed a three-node architecture: one control-plane node and two workers, with custom VPCs, subnets, and IAM configuration.
• Automated containerd installation, kubeadm initialization, Calico CNI deployment, and certificate management with Ansible, Python, and Bash.
• Built modular system configuration, security-hardening, and worker-join components.