DevOps engineer on a multi-service cloud platform, in an aerospace environment.

• AWS costs cut by 65 to 75 % depending on the service: audit of real provisioning service by service, then a reduction in two one-week steps with CloudWatch validation between them, with zero production incidents.
• Securing the multi-service platform (EKS, ECS, Lambda, RDS · Aurora, S3, CloudFront, Cognito, SNS, SageMaker): WAF and Shield hardening, centralised secrets management (Secrets Manager, KMS), security and performance alarms.
• Design of the data ingestion pipelines: AWS Glue jobs scheduled by EventBridge, S3 storage, secrets managed by Secrets Manager.
• Development of an AI-augmented search bar (RAG) on Google Cloud Platform and Gemini.
• Quality and security industrialisation of the pipelines: Checkmarx (SAST), SonarQube and Cypress built into CI/CD, detected vulnerabilities fixed, artefacts managed in JFrog Artifactory, Jenkins jobs parallelised.
• Database migration with KMS encryption, Terraform refactoring, CloudFormation deployments, several production releases delivered.
• Application support and tooling in Python · FastAPI (back end) and React (front end).

Environment : AWS, GCP, Gemini, Terraform, CloudFormation, Checkmarx, SonarQube, Cypress, Jenkins, JFrog Artifactory, React, Python, FastAPI.

DevOps engineer, full ownership of the project’s cloud environments.

• End-to-end ownership of the Azure environments: development, test and a fully isolated sandbox, set up from day one of the project.
• Definition and enforcement of network security rules across all environments.
• Set-up of a FinOps alerting system for real-time cloud cost tracking, and centralised secrets management.

Environment : Microsoft Azure, Azure DevOps, CI/CD.

Software engineer, Agile SAFe framework. Critical apps for tracking, maintaining and repairing agricultural machinery.

• Technical debt reduced by about 80 % through refactoring and a GitLab repository clean-up script that removed more than 800 obsolete branches.
• Cordova to Capacitor technology migration, with a detailed impact analysis on the plugins and the app.
• Mobile CI/CD chain set up (Jenkins for iOS builds, Azure DevOps), full release cycle management and six quarterly releases on Google Play and the App Store.
• Delivery of a Jenkins pipeline POC to Azure and TestFlight, with turnkey documentation for the client (variables, secured certificates, deployment procedures).
• Integration of Firebase Crashlytics, Matomo, iOS and Android push notifications and Google Maps.

Environment : Ionic, Angular, Capacitor, Cordova, Java Spring, Jenkins, Azure DevOps, GitLab, Firebase, TestFlight.

Software engineer, Agile SAFe framework. An international project of over one hundred people between France and India.

• Major technology migration of the monitoring web application, from AngularJS to Angular 16.
• Business features developed with the France and India teams, production maintenance, legacy code refactoring and unit tests.

Environment : Angular, AngularJS, Java Spring, Jenkins, GitLab, Jira.

COVID-19 context. End-to-end responsibility: design, development, deployment and production release.

• Design of the complete AWS infrastructure: back end on Beanstalk, front end on S3 and CloudFront, RDS database with auto-scaling, replication and automated backups.
• Development of an SEO-optimised serverless PWA (Angular Universal, Node.js, AWS Lambda, API Gateway), Docker containerisation and Azure DevOps CI/CD.
• Stripe payment module, real-time chat over WebSockets, releases on Google Play and the App Store, cost control with AWS Budgets.

Environment : Angular, Ionic, .NET Core · C#, AWS, Docker, Azure DevOps, Stripe.

Designed, built and run alone, in 24/7 production.

• Platform engineering: an internal platform whose consumers are AI agents. The constraint is that of a team of developers: they must be able to act without me. Hence an automatically generated system state, a catalogue of reusable capabilities, explicit usage rules and test benches.
• Platform tooling: Python micro-services (Flask, FastAPI) in containers, an MCP connector exposing capabilities to an external model, webhook-driven generation workflows, multi-channel ingestion (text, voice, image, PDF, URL, email).
• Privacy by design: two-lane deterministic routing, public data to a cloud LLM, data marked private kept in place, with no model or external API at all.
• Security: encrypted private access, allow-lists, a secret guard before anything leaves, encrypted off-site backups, a written recovery procedure.
• SRE-style operations and LLM FinOps: automatic restarts, failure alerting armed and verified in production, diagnosis of a silent failure. A GPU workload moved to an on-demand service, at a measured cost of $0.041 per run, about $6 a month.
• Delivery chain: multi-architecture continuous integration, infrastructure as code (OpenTofu) and GitOps for Kubernetes, written and bench-tested; runbooks and blameless post-mortems.

Environment : Docker, n8n, OpenTofu, Python, Claude Agent SDK, MCP, Git.