AI and process automation
Assistants, agentic workflows, RAG and LLM integrations connected to the tools already used by the company.
- Controlled access to data and actions
- Quality evaluation before production
- Cost and human oversight under control
SOFTWARE FOR BUSINESS · JAVA / KOTLIN · PRACTICAL AI
For companies in Poland, I design custom applications, backends and AI automation. I turn over 10 years of JVM experience into solutions that can be safely deployed and maintained.
01 / SERVICES
I select technology for the process, risk and expected outcome. I do not force AI where a simpler system works better.
Assistants, agentic workflows, RAG and LLM integrations connected to the tools already used by the company.
Web products and internal tools designed around the user's real process.
Kotlin · React · PostgreSQLAPIs, distributed systems, JVM modernization and secure service integration.
Java · Spring · Ktor · CloudA review of risk, performance and change cost, ending with a plan that can be executed.
Architecture · Delivery · Quality02 / AI IN PRACTICE
A good AI system knows the right context, uses permitted tools and makes quality measurable. The model is one component, not the entire product.
Let us analyze your processData, business rules and a clear task objective.
APIs and actions that let AI perform controlled work.
Evaluation, observability, permissions and cost limits.
Approval wherever an error has real consequences.
03 / EXPERTISE
I combine enterprise experience with the speed of a small team. I own the technical decisions, code, deployment and continued operation of the system.
years of commercial work on web applications and enterprise systems
Backends, integrations, microservices and systems that must remain stable.
AI connected to data, tools, evaluation and a real process.
Cloud, containers, monitoring and documentation are part of the solution.
Check the technical owner, public profile and working booking system instead of relying only on claims.
I select the scope for the process, data and risk, without hiding infrastructure behind a marketing diagram.
04 / COLLABORATION
The problem, users, data and constraints before choosing technology.
Short iterations, working increments and regular direction checks.
Deployment, observability, documentation and knowledge transfer.
STAWIARI · FOUNDER / SENIOR JAVA DEVELOPER
I do not pass the project between sales and an anonymous team. You speak with the person who designs the solution and is accountable for delivering it.
05 / SHORT AND SPECIFIC
No. AI only makes sense when it improves a specific process. If simpler automation is cheaper and more reliable, I choose the simpler solution.
With a short conversation about the problem, users, data and constraints. Its outcome is a sensible next step, not a sales presentation.
Yes. An audit, technical prototype or first working increment often reduces risk before a larger investment.
06 / FIRST STEP
Choose an available time. We will discuss the problem, constraints and a sensible next step, without a sales deck.
Available times will appear here.
Save the event to your calendar. Keep the management link in case your plans change.
07 / CONTACT