Cloud architecture
Multi-cloud and disaster recovery designs: AWS as the primary site with Azure for recovery, active/passive across zones, Kubernetes at scale, a 99.999% availability target.
Freelance cloud architect · available January 2027
An architect on your team, with cost as a design constraint.
I join teams for architecture, migration and performance work on AWS, Azure and Google Cloud. Every design choice comes with its monthly cost.
What I do
Migrations and performance
Rewriting the parts that cost the most, with the legacy system as the reference for every test until the numbers match.
Infrastructure as code and security
Terraform, Terragrunt, Bicep and Ansible, with plan reviews, automated tests, policy as code and drift detection. Identity with Entra ID, OAuth/OIDC and SAML.
Cost from day one
FinOps at the design stage rather than after the bill arrives: each option is priced before it is built.
Case · EDF Power Solutions
A simulation engine rewritten from C# to Rust.
99%
faster: from about 7 h 30 to under 30 seconds per run
−50%
cloud costs, approximately
÷40
peak memory at the results stage
Data extracted once to object storage as frozen columnar files. Short single-purpose steps that talk through storage, not to each other. Independent slices run in parallel on short-lived serverless functions on ARM, sized step by step.
Case · Energy, real time
A real-time IoT platform built to scale and update without downtime.
1M+
real-time IoT sensors
< 3 min
to recover on Azure if AWS goes down across zones and regions
NIS2
conformant security framework, now the reference for adjacent systems
An elastic ingestion and processing pipeline on AWS, designed to keep data intact and available at large scale, with the whole infrastructure replicated 1:1 on Azure for disaster recovery. FinOps and DevOps were part of the design from the start, with infrastructure as code in Terraform, containers on Docker and Kubernetes, and CI/CD. Elasticity and zero interruption were bought with a much larger operational and security surface, which made NIS2 conformance load-bearing.
Case · EDF Power Solutions
A file ingestion chain built for very high throughput and delivered to production.
50 TB
of data a day: the volume the chain was sized for
In the same engagement, a high-performance computing platform on AWS, designed and implemented as code: managed SLURM, FSx for Lustre, MPI and tightly coupled parallel jobs on an auto-scaled EC2 cluster.
Governance · ISALYS Group
Architecture standards built into the delivery contract.
3
countries under one architecture method
5
client contexts under one development governance charter
Technical director of the engineering entity of a group of about 350 people. Standards, reference integration patterns and a development governance charter with three stated criteria: explicitness, simplicity, maintainability. ADRs made a habit rather than a mandate: teams got the format, the charts and protected time to write them. C4 introduced as the shared vocabulary across domains.
Decision · ISALYS Group
A two-tier LLM runtime instead of one vendor.
Managed frontier models for reasoning-heavy features, self-hosted lightweight Mistral models where data residency or per-token cost dominates. The trade-off: two runtimes and two security baselines, against lock-in and a token bill nobody could attribute. Cost per feature became a design input, and the reach of each data-processing agreement is decided at the boundary, per capability.
1 / 5
Stack
RustC / C++PythonAWSAzureGoogle CloudKubernetesPostgreSQLTerraformTerragruntBicepAnsibleEntra IDHPC
Next step
Tell me about the mission.
Pick a slot and tell me the context, dates and team. Prefer writing? contact@mathieukerbel.com