Industrialising AI on a group-wide Azure platform
Turning AI pilots into production services on a shared Azure platform, with automated delivery and an API management layer in front of every model.
We engineer the systems large organisations run on.
Cloud, DevOps, software and AI engineering by senior engineers, from architecture to production.
14 years
in enterprise IT and operations
Insurance · Banking · Assistance
regulated sectors we know
−60 %
cloud cost on a critical API platform
≈ 100 apps
migrated off obsolete systems
01 Services
Four disciplines, one engineering team.
We work where infrastructure, delivery and software meet. Most of our engagements span more than one of these.
- 01
Cloud
Landing zones, migrations and hybrid platforms on Azure, AWS, Scaleway or your own datacentre. Built as code, governed from day one, with the bill under control.
- Landing zones
- Migration programmes
- Private and hybrid platforms
- 02
DevOps
CI/CD platforms, software factories, infrastructure as code and observability. Fewer manual steps, faster releases, and incidents you see before your users do.
- CI/CD platforms
- Software factory
- Infrastructure and configuration as code
- 03
Software engineering
API-first architecture, microservices, API gateways and event-driven integration. Software that is modular, tested and documented, built by engineers who also run it.
- API-first architecture
- Microservices
- API management
- 04
AI & Data
AI platforms, LLM integration, retrieval-augmented assistants and data pipelines, industrialised with the same rigour as the rest of your systems, and ready for the EU AI Act.
- AI platforms
- LLM gateways
- Retrieval-augmented assistants
02 Selected projects
Our team has delivered for insurers, banks and service groups.
Client names stay confidential. The engineering does not.
Migrating a legacy estate and bringing Kubernetes on-premise
An obsolescence programme covering about a hundred applications, plus an on-premise Kubernetes platform aligned with the group's Azure services.
≈ 100applications migrated off obsolete systems
Running a critical CRM API platform on hybrid cloud
Operating and automating the APIs behind the insurer's customer relationship core, while cutting their cloud costs by more than half.
24APIs operated and monitored
03 How we work
Pick the engagement that fits the problem.
01
Consulting & audit
Architecture reviews, cloud and FinOps audits, platform roadmaps. A clear diagnosis and a plan your teams can execute.
02
Team extension
Senior engineers embedded in your squads, following your rituals and tooling, with knowledge transfer built in.
03
Delivery projects
A defined scope, a committed outcome: a landing zone, a migration, a CI/CD platform, an AI service in production.
04
Managed run
We operate what we build: monitoring, incident response, upgrades and cost control, under agreed service levels.
04 Technology-agnostic
Your stack, not ours.
Public cloud, private cloud or both. Kubernetes or serverless. We have shipped on most of the platforms large organisations rely on, and we pick tools for your constraints, not our habits.
05 Our solutions
Djinn, sovereign AI services for European enterprises.
We build what we recommend. Djinn brings ready-to-use AI services together behind one API, hosted in France, with every decision traceable.
06 Values
How we engineer.
01
Engineering rigour
Everything as code, every decision written down. What we build can be reviewed, reproduced and handed over.
02
Sovereign by default
European hosting and data residency whenever the use case allows, and a clear view of the trade-offs when it does not.
03
Technology-agnostic
We adapt to your platforms and contracts. Recommendations follow your constraints, never a vendor quota.
04
Measured outcomes
Cost, lead time, availability: we agree on the numbers up front and report on them.
05
Leave teams autonomous
Documentation, pairing and workshops are part of the job. Success is when you no longer need us for it.
Have a platform to build, migrate or fix?
Tell us about it. An engineer, not a salesperson, will answer within two working days.
Start the conversation →