Planning a move to AWS, a switch to Kubernetes, or simply want to stop deploying by hand? I design, migrate and automate cloud infrastructure for SMEs and startups: AWS, GCP or Azure architecture, EKS clusters, infrastructure as code with Terraform, CI/CD pipelines and monitoring. The goal is simple: infrastructure that is reliable, reproducible and whose costs stay under control.
Freelance DevOps consultant · ESIEA engineer · 10+ years in cloud and infrastructure engineering · AWS Certified Solutions Architect · Kubernetes CKA.
No commitment · one technical point of contact · on quote
Verified certifications
What I take on
An engagement can cover one of these topics or all of them. Either way, what gets delivered runs in production and stays maintainable by your team.
Architectures sized for your real workload: networking (VPCs, subnets, peering), compute, managed databases, storage and security. Every choice is documented and justified, including when the right answer is to keep things simpler.
Migrating on-premise servers or VMs to the cloud, containerizing applications and moving them onto Kubernetes. Databases included, with a cutover and rollback plan agreed well before the big day.
Building and taking over managed clusters, including EKS Auto Mode, which hands node management over to AWS. Ingress, autoscaling, secrets management, RBAC and version upgrades without downtime.
Your entire infrastructure described in versioned code with Terraform or AWS CDK: reusable modules, identical environments from staging to production, and code review before every change.
Automated build, test and deployment with GitLab CI, GitHub Actions or Jenkins. Runners hosted on your own infrastructure when needed, progressive rollouts and one-click rollback.
Metrics, logs and alerts that are actually useful: Prometheus, Grafana, CloudWatch or Datadog depending on your context. Readable dashboards, and alerts that only wake someone up when it is justified.
Account and project structure, naming and tagging conventions, security policies and a catalog of standardized services. A clear framework so every team can ship fast without reinventing the wheel.
One directory to access the cloud console, the clusters and your internal tools: SSO through Entra ID, Amazon Cognito or IAM Identity Center. Fewer accounts to manage, and leavers handled in a single action.
Sizing, cost tags and budgets built into the architecture. If your bill has already gotten out of hand, the FinOps audit tackles that first.
Who it is for
SMEs, startups and tech teams without a dedicated infrastructure engineer, or with a team that needs extra hands for a while.
Hardware to replace, a hosting contract coming to an end, a server room to close: the right time to plan an AWS cloud migration instead of renewing the same setup.
Manual releases, a single person who knows how everything works, environments drifting apart. Infrastructure as code and a CI/CD pipeline fix this at the root.
You want an independent opinion before committing, or a Kubernetes consultant to build a clean EKS cluster from the start rather than fixing it later.
Accounts created as you went, overly broad permissions, resources with no owner, no shared monitoring. Resetting governance and access prevents the next incident or the next surprise bill.
Method
Four steps, each one signed off by you before moving on. Nothing changes in production without a written plan.
Inventory of servers, applications, network flows, dependencies and pain points, with read-only access. We identify what needs to move, what can wait and where the risks are.
Target architecture, choice of managed services, a migration strategy application by application and an estimate of running costs. You approve before anything gets built.
Infrastructure written in Terraform or CDK, CI/CD pipelines and monitoring, then a progressive cutover of applications with tests and a rollback plan at every step.
Operations documentation, incident runbooks and knowledge-transfer sessions with your team. You stay autonomous, with the option of ongoing support afterwards if you want it.
Deliverables
Everything stays with you: the code, the access and the documentation. No lock-in to a proprietary tool or to my presence.
References
A selection of infrastructure and DevOps engagements for industrial groups, digital businesses and a video game publisher.
Chemicals · Industry
Migration of CI/CD runners to an EKS cluster. Identity federation (Cognito / Entra ID) for centralized access to tools.
Automotive supplier
Infrastructure-as-code design integrating managed services. Centralized access control on a critical environment.
Water treatment
Migration of an on-premise cluster to Amazon EKS, database included.
Textile · Distribution
A cloud governance and best-practice framework.
Online platform
Design of a service catalog.
Video games · SRE / DevOps engineer
In the debug & profiling team: handling production crashes, monitoring servers and updating developer tooling through pipelines on Kubernetes.
Digital services
Monitoring overhaul and management of the on-premise infrastructure.
Pricing
Migrating three servers and taking over a multi-team Kubernetes platform are not the same scope. The quote is written once the need is understood, not before.
We talk over video about your current infrastructure and your goal: migration, Kubernetes, automation, reliability. You leave with a candid technical opinion, first leads and, if it makes sense, a costed engagement proposal.
Offering
Understand and reduce your AWS, GCP or Azure bill.
Kubernetes
AWS runs the nodes, scaling, networking and storage.
AWS architecture
LiteLLM on AWS, costed line by line.
Infra & AI
Claude Code and Amazon Bedrock on infrastructure you control.
All three providers cover what an SME needs. The choice mostly depends on your context: your team's skills, the tools already in place (a Microsoft 365 and Entra ID estate often points to Azure, heavy BigQuery use to GCP), the managed services you need and your data residency constraints. AWS still has the broadest catalog and it is my main area of expertise, but I recommend the cloud that suits you, not the one I prefer.
The new infrastructure is built and tested alongside the existing one. Data is synchronized continuously where possible (database replication, incremental file copies), then the cutover happens application by application, in a window chosen with you, with a rollback plan written in advance. How long each application is unavailable depends on the application and its database; it is estimated during the design phase, before anything is committed.
Not always. For a handful of simple applications, services such as ECS, Cloud Run, App Runner or a well-automated VM are enough and cheaper to run. Kubernetes starts to pay off when you have several containerized services, several teams or real scaling needs. Managed offerings like EKS Auto Mode cut the operational burden a lot. As a CKA-certified Kubernetes consultant, I will tell you plainly whether it is the right tool for you.
Your infrastructure becomes code versioned in Git: every change is reviewed before it is applied, the history shows who changed what and why, and a full environment can be recreated identically. It removes hand-made configuration in the console, makes security audits easier and speeds up recovery after an incident considerably. Terraform works on AWS, GCP and Azure; AWS CDK is a good option if you are all-in on AWS and your team prefers TypeScript or Python.
Yes, that is the most common setup. I plug into your tools (Git, ticketing, chat) and your rituals, pair with your developers or operations team and document as I go. The goal is for your team to be autonomous on everything that was built. I can also work on my own if you do not have infrastructure skills in-house yet.
Yes, engagements are run remotely, in English or French. Cloud infrastructure is administered remotely anyway, and we work together over video, in your team chat and in your code repositories. Access is granted with the minimum permissions needed and revoked at the end of the engagement.
A move to AWS, a Kubernetes cluster, a CI/CD pipeline or tidying up what you already have: describe your situation in a few lines and we set up 30 minutes to discuss it. No commitment, with the engineer who will actually do the work.
Is your bill the priority? Take a look at the FinOps audit instead, or browse all our offerings.
Book my free diagnosticFree · 30 minutes · a candid technical opinion