Four engagements, described honestly.

A Romanian bank had no online banking product for corporate customers at all. We took it from concept to production on Azure.

The situation

No previous solution existed — this was a greenfield build of the bank's entire online banking platform for corporate entities, not a modernisation of something already running. The bank had an internal development team, but no established CI/CD practice and limited Kubernetes exposure.

What we did

We owned the architecture end to end: designed the target infrastructure, proposed and agreed the final design with the client and every stakeholder, and implemented the whole thing across three environments on Azure.

The platform runs on AKS with Azure-native components throughout, and a custom ElasticSearch/ELK stack for logging and monitoring. Redundancy, backups and resilience were part of the design from day one rather than a later hardening pass — including an active-standby high-availability topology where the standby replica is kept permanently in sync as an exact copy of the primary.

Alongside the build we coordinated and guided the bank's own development team, bringing them up to speed on CI/CD and modern DevOps and security practice.

Outcome

Production go-live, with the bank's team operating the platform and working within the delivery practices established during the engagement.

• Microsoft Azure
• AKS / Kubernetes
• ElasticSearch / ELK
• Active-standby HA
• CI/CD
• Three-environment topology

A Romanian government agency needed a national portal for how citizens deal with energy providers — on their own hardware, run by a team new to Kubernetes.

The situation

The agency was building a national portal to streamline how Romanian citizens interact with energy providers. Public cloud was not the answer here: the platform had to run on the agency's existing on-premises, virtualised VMware ESXi estate. Their internal IT team had no prior cloud or Kubernetes experience.

What we did

We designed a modern self-hosted Kubernetes infrastructure, working side by side with the agency's internal IT team rather than around them — which mattered, because they had to own it afterwards.

Multiple Kubernetes clusters were architected and deployed as separate replica environments — development, QA and production — across the on-premises VMware estate, with redundancy, high availability and very low latency as explicit design targets. We worked with the development team throughout to establish CI/CD and DevSecOps practice as the portal was built.

Outcome

A secure, redundant national platform, and — the part that took the most deliberate effort — an internal team that could confidently operate and maintain Kubernetes infrastructure they had never worked with before. Knowledge transfer was run as a structured hand-off, not a final-week document dump.

• Self-hosted Kubernetes
• VMware vSphere / ESXi
• Dev / QA / production clusters
• High availability
• CI/CD
• Structured hand-off

A global healthcare and medical services provider needed its change and development processes actually enforced across a worldwide Azure footprint.

The situation

The client operates globally, with development teams spread across their worldwide operations and a strict set of change and development processes governing their Azure estate. Processes existed; consistent enforcement across that many teams was the harder problem.

We were engaged through a managed-services provider. Within months the role became Team Lead, and at the request of the client's senior management the work moved to being embedded directly within their internal engineering team.

What we did

We owned the security of the Azure environments: reviewing and approving change requests for compliance, continuously auditing existing infrastructure, and working directly with development teams across the client's global operations. That ran alongside the vulnerability management and SOC teams to keep non-compliances from accumulating.

The most durable piece was automation. We proposed, designed and implemented automated dependency scanning with Dependabot and automated container image patching with Copacetic — closing a significant standing risk from unmaintained Docker images that manual review was never going to keep up with.

Outcome

Unmaintained container images stopped being an open risk, and compliance review shifted from a queue of manual approvals toward controls that hold on their own.

• Microsoft Azure
• Dependabot
• Copacetic
• Container image patching
• Change governance
• SOC collaboration

A US fintech serving major banks was carrying a legacy Azure product with structural problems. It was rebuilt on new AWS infrastructure.

The situation

The client runs an AI-based mortgage platform used by major US banks. The existing Azure product had accumulated deficiencies serious enough that incremental repair was no longer the cheaper option — the infrastructure needed replacing rather than patching.

What we did

We worked directly with the CTO, the senior product owner and the Head of DevOps, and designed a large part of the internal infrastructure for the replacement AWS platform.

All of the Infrastructure-as-Code behind the new environments was authored in OpenTofu/Terraform — a modular design intended to be extended by the client's own engineers across multiple environments, rather than a one-off deployment script that ossifies the moment the consultant leaves.

Outcome

A new AWS platform replacing the legacy product, with the infrastructure defined as reviewable, modular code the client's team owns.

• Amazon AWS
• OpenTofu / Terraform
• Modular IaC
• Multi-environment
• Platform re-architecture