Enterprise Systems Design and Development
The systems in the middle of the company
Order management, billing, provisioning, the integration layer that ties them together. These are the systems where a bad design costs money every day for years. We design and build them, and we have been doing it for a long time.
Most of our work is in Java, on Kafka, Cassandra, relational databases and Kubernetes. We are happy to work in your stack if it is a reasonable one.
Contact usHow a project goes
We learn your business first
What the system is for, who uses it, what goes wrong today. We read the existing code and look at the existing infrastructure before we propose anything.
Then we design
A written design covering data, scaling, security and integrations, with the trade-offs spelled out. You should be able to hand it to another vendor and get the same system.
Short iterations
Working software every couple of weeks, demoed to the people who will use it. Course corrections are cheap early and expensive late, so we ask for feedback early.
Built to grow sideways
Stateless services, partitioned data, queues between components. When load doubles, you add machines rather than rewriting.
Integration is part of the job
A new system has to talk to the old ones: databases, SOAP services, message queues, nightly files. We treat those interfaces as first-class work, not an afterthought.
Testing, including the ugly cases
Unit and integration tests, load tests at realistic volumes, and failure tests: kill a node, fill a disk, lose a broker. The system should survive all of it before it meets a customer.
We stay after go-live
We help with the deployment, we are on call for the first weeks, and we offer ongoing maintenance if you want it. We do not disappear when the invoice is paid.
What you end up with
Less manual work
Steps that people do by hand today, in spreadsheets or by re-keying data, become part of the system.
Room to grow
A system designed for ten times today's volume, so the next growth spurt is a capacity change rather than a rewrite.
One version of the truth
Data that is stored once, in a model people can query, instead of copied between departments and argued about.
Systems that talk to each other
When sales, operations and finance see the same records at the same time, most of the meetings about "whose number is right" go away.
Who you would be working with
Engineers with 15 or more years each on systems like these, mostly in telecom, where the volumes are large and the tolerance for errors is zero. We are a small team, so the people who design the system are the people who write it.
Write to us about your project. If you need an opinion rather than a team, our enterprise systems consultancy is the shorter engagement.