I modernise critical systems without taking them down.

If any of these sound familiar, I can probably help.

• An old .NET application that has to keep running?
• A .NET Framework to .NET 8 migration with no appetite for a big bang?
• A legacy system you inherited and nobody wants to touch?
• Hard problems that only show up in production?
• Old systems that need to talk to modern APIs and services?
• A team that needs a Tech Lead to own the architecture decisions?
• A critical platform that must be modernised without stopping operations?
• Cloud or AI to bring into an existing system without rewriting it?

15+ years

Software engineering, most of it on systems that were already in production.

.NET

Modernising .NET Framework, WebForms, DNN and monoliths to .NET 8.

Azure + AWS

Cloud, containers and distributed systems in regulated environments.

Production

Critical systems, high accountability, changes with no maintenance window.

Sectors: energy · banking · public sector · telecoms · mobility

Three real problems, solved in production.

energy · payments · aws

Cards vanishing right after the customer set up an MB WAY mandate

Problem

The app confirmed the mandate, then showed no card at all. Intermittent, and only in production.

Diagnosis

Reading OpenSearch logs against the token documents in MongoDB showed the cause was not in our code. The follow-up inquiry went out roughly 750 ms after creation and caught the payment provider mid-sync, which answered 404. A race condition between two systems, invisible in test environments.

Solution

A pre-check on token insertion and a compound index that cleared the duplicates already stored. On .NET 8 Lambdas, ECS and DocumentDB, without changing the flow the customer sees.

Outcome

Cards stopped disappearing and duplicates were blocked at the source, with no rewrite of the integration and no downtime.

public sector · document management · dnn

Two decades of archive moved from cloud to on-prem without breaking a single link

Problem

A document library published since 2004 on a DNN portal, with public URLs built on filetickets derived from the FileId. The files had to leave blob storage and land on the internal fileserver.

Diagnosis

A different FileId meant a different ticket, so any link indexed over the years would die. And the source storage was shared with production, where nothing could be deleted. Moving was risky; copying while preserving identity was not.

Solution

Copy keeping the FileId instead of moving, and validate ticket by ticket in development before touching production.

Outcome

Twenty years of documents migrated with no broken links and no portal downtime.

energy · workflows · k2

Work authorisations getting stuck mid-process

Problem

K2 workflow instances stuck halfway, on a platform operations use every day and nobody wants to touch anymore.

Diagnosis

Blocks on variable initialisation and mappings reading the field off the wrong table. Unglamorous work that required understanding the platform before changing it.

Solution

Fixed the mappings and wrote idempotent SQL migrations with rollback to repair the tasks already created.

Outcome

Processes unblocked without replacing the platform or interrupting the people who depend on it.

Clients are under NDA, so I describe the sector and the scale rather than the name. On a call I can be more specific.

Not everything needs a rewrite.

• Not every legacy system needs replacing.
• Not everything needs to become microservices.
• Not every system needs Kubernetes.
• Not every application needs AI.
• First I identify the risk; then we decide what is worth changing.
• Incremental modernisation is almost always safer than a big bang.

How I can help.