Backend
Java services that carry the load
REST APIs, batch processing, messaging and persistence with Java and Spring Boot, on systems that book, calculate and settle and have to be right.
Zagreb Europe
Software that fits the task
A company website, a data migration, a platform in daily production use. The service is scaled to the problem, not the other way round.
- Java Spring Boot
- Angular
- Azure Kubernetes
- Oracle SQL Server DB2
- Kafka Elasticsearch
Services offered
Small jobs and large builds
Both are welcome, and both are quoted the same way: a written scope with the effort the task actually needs.
Frontend
The screens on top of them
Angular and TypeScript for business applications, plain HTML and CSS for company websites. A small static site or a compact service for orders is welcome work; the weight of the practice sits on the large backend systems.
Deployment
Cloud, or a single server
Software written to run as a container on Azure and Kubernetes, or on a single server when that is enough. We build the application, package it and set up the deployment path; the infrastructure itself stays with your team or your provider.
Architecture
Where the boundaries run, and why
Service boundaries, data models, API contracts and the reasoning behind them, written down so the next developer can follow it and revisited when reality disagrees with it.
AI integration
Language models inside business software
Classification, extraction and search over your own documents, built with validation, confidence thresholds and a person in the loop, so nothing reaches a business system unchecked.
Requirements
The specification before the code
Reviewing what exists, naming the gaps and contradictions, and writing something the business side and the developers can both sign. The part that gets expensive later when it is missing.
Engineering capabilities
Technology, architecture, value
The stack the work is built on. Chosen for what a system should do, not for what is new.
Cloud and platform
Systems designed for load, scaling and operation on a cloud platform.
- Azure
- Kubernetes
- Docker
- Helm
- CI/CD
- Grafana
- Monitoring
Software engineering
Robust backend services and modern front ends.
- Java
- Spring Boot
- Spring Batch
- Angular
- REST APIs
- Microservices
Data and integration
Systems kept in step, and the data in them put to use.
- Kafka
- IBM MQ
- SQL Server
- PostgreSQL
- Elasticsearch
- Event streaming
- Data modelling
Applied AI
Language models used where their benefit can be checked.
- LLM integration
- RAG
- MCP
- Document classification
- AI workflows
- Automation
Quality and security
Tests, traceability and security requirements from the first increment.
- Testing
- Code quality
- OAuth2
- SonarQube
- Security
- Observability
- Logging
- Compliance
Project examples
Six setups, and the decision each one turns on
Examples from the sectors this work comes from, from a corporate system down to the website of a small company. Each describes the setup and the question it hangs on.
Insurance · Cloud platform
A microservice landscape with a cross-cutting layer
Dozens of small services on a cloud platform, and underneath them a layer holding everything they all need: a BPM engine for the long-running business processes, a central authorisation server, a replicated in-memory session store so any instance can answer any request, and shared libraries for logging and error handling. The cost of this shape is that cross-cutting layer. The benefit is that the individual services stay small enough to be replaced.
- Stack
- Azure
- Spring Boot
- Camunda BPM
Document-heavy process · AI
A model that proposes and an architecture that decides
Incoming documents pass through a chain: classification, extraction into a validated schema, indexing for search. The model only proposes. The architecture decides what happens next — anything below a confidence threshold goes into a review queue for a person instead of into the business system. Each step is its own service, so the model can be exchanged without touching the process around it.
- Stack
- Azure
- LLM API
- Elasticsearch
Retail · Store accounting
New services beside a mainframe application
A host application grown over decades, with a very large database on the mainframe, remains the leading system and is not going to be replaced. New capability is added next to it as separate services that read the same data and publish their results to the neighbouring systems over REST and messaging. The architectural question is not how to retire the host, but where the boundary between host and service runs and which side owns which data.
- Stack
- Java
- Spring Boot
- Mainframe DB2
Public administration · Batch
Scheduled processing with the rules held outside the code
A system whose work happens in scheduled runs rather than in requests: jobs with restart points and per-record error handling, so a run that fails part-way resumes instead of starting again. The decisions that change with legislation live in a rule engine rather than in the code, so they can change without a release, and adapters for file transfer, mail and SOAP face the neighbouring systems that cannot be changed at all. Throughput and restartability matter more than latency.
- Stack
- Java EE
- Batch processing
- Rule engine
Financial services · Advisory application
A web front end over a system that cannot be touched
An established back-office system supplies the master data and executes the transactions, and it cannot be modified. In front of it sits a backend that holds the session, merges data from several sources and exposes one narrow REST interface to an Angular application. The work is not in the screens. It is in that middle layer: timeouts, duplicate answers and idempotency, so that a transaction the user triggered once takes effect exactly once.
- Stack
- Angular
- REST/JSON
- Legacy system integration
Small company · Website
A company website with no CMS to maintain
A handful of pages in several languages, written as plain HTML and CSS and delivered as static files: fast on a phone, no cookie banner because there is nothing to consent to, and hostable almost anywhere for a few euros a month. Design drafts along the owner's own ideas come early, before any technology is chosen. Later changes are either handed over as a simple editing path or done on call.
- Stack
- HTML / CSS / JS
- Static hosting
- Multilingual
Describe the task
A short description of the problem is enough for a first assessment and an honest answer on scope — including when the task is smaller than expected.
German and English are the everyday working languages; enquiries in any other language are equally welcome and answered in the one you wrote in.