Software architect · Advisor · Co-founder, Arpeggia Studio

I help companies decide what to build, before the AI decides for them.

For 22 years I've built systems from the inside: as a developer, team lead and lead architect for logistics, public sector, insurance and telco. Today I help CTOs and founders turn unclear ideas into buildable products and make technology decisions that still make sense in five years. I'm a big fan of AI: I use it every day and teach teams how to use it wisely. Former engineer. Still reads the code.

  • 22years designing & building software
  • 1stO'Reilly Architectural Katas, Winter 2024
  • 14countries served by one platform I architected
  • 100+production deployments in my engineering years
  • 1technical book written for Packt Publishing
01 / Concepts and approach

How I approach architecture.

Three things fit together here. Responsible Software Architecture says what a good design has to take care of. The six-step process is how I get there on a real project. Architectural Kata 2.0 is where people practise it.

  1. WhyResponsible Software ArchitectureFour pillars a design has to answer to
  2. HowSix-step design processFrom requirements to team boundaries
  3. PracticeArchitectural Kata 2.0A workshop to rehearse the process

Concept, co-created in 2024

Responsible Software Architecture

The problem: architecture decisions often focus on the technical side alone and leave out people, purpose and long-term impact.

What it is: a framework of four pillars that every significant design decision should be checked against.

Who: I created it together with Sebastian Dąbkowski and Wojciech Kasa, my co-founders at Arpeggia Studio.

responsiblesoftwarearchitecture.com ↗

Responsible Organization

Transparent decision-making, ethical frameworks and diverse, healthy teams that build the system.

Responsible Business

Technology choices aligned with sustainable business practice and long-term value for all stakeholders.

Responsible Code

Maintainable, secure and accessible code that follows good engineering practice.

Green Software

Energy-efficient applications that keep their environmental impact low.

Process I developed, 2024 to 2026

Six steps from a brief to a buildable design

The problem: teams jump from a brief straight to technology choices and discover missing requirements, risks and ownership questions too late.

What it is: six steps, each ending with an artifact the next step builds on. The steps themselves rely on established techniques by other authors, credited in each step. My contribution is the sequence and how the steps connect.

See it applied: the examples come from MonitorMe, the design that won the O'Reilly Architectural Katas. See the case below ↓

The steps interlock, like the faces of a Rubik's cube: turn one and the others move with it. When a later step brings new information, I go back and revise the earlier ones.

Step 01 of 06

Functional requirements

What the system must do, for whom, and why the business cares. Before boxes and arrows, I want the key flows and the outcomes they pay for.

I ask
Who are the actors? Which flows make money or save it? What happens if this flow is late or wrong?
Artifact
Event Storming board, actor and flow map, prioritised capabilities
Built on
Event Storming
Example
MonitorMe: collect vital signs, show them at the nurse station, alert staff on anomalies, keep 24 hours of history.

Workshop format, run with Sebastian Dąbkowski and Wojciech Kasa

Architectural Kata 2.0

The problem: architects rarely get to practise decisions without the risk of a live project. What it is: a one-day workshop where teams take the same business problem through the six steps, use AI agents that question their choices, and defend the result in front of a demanding client.

02 / Career: the whole path, already walked

Been there, built that.

Today I advise at board level. Over 22 years I held every job below it, down to writing the first line of code.

Pick a level, or click a box in the diagram, to zoom into that stage of my career

Business2025 →C4: Business context

Advisor & Co-founder, Arpeggia Studio

Where I am today. I sit with founders, CTOs and boards before the money is spent: what to build, which vendor to trust, whether to rescue or replace. With AI as a tool, not an excuse.

  • Product discovery
  • Tech risk
  • AI agents
  • KATA workshops
03 / Case study

First place at O'Reilly Architectural Katas.

Winter 2024. An international competition judged on resilience, security, scalability and, above all, the quality of the reasoning behind every decision. Team BluzBrothers built the whole proposal in three evenings.

Vital-signdevices ×NDevicegatewayFailed deviceisolated, rest keeps goingEvent brokervitals.* streamsasync · fault-tolerantNurse stationlive view ≤ 1 sAnomaly alertsmobile + stationVitals store24 h historyMyMedicalDatasnapshots
StayHealthy · MonitorMe: simplified container viewStyle: event-driven
1stplace
O'Reilly Architectural Katas, Winter 2024
Problem
Real-time patient monitoring for hospitals: vital signs from many devices, intelligent alerting, integration with hospital systems. Life-critical.
Drivers
≤ 1 s average to display vitals · keeps monitoring when a device fails · high availability
Decision
Event-driven architecture, chosen for performance, availability and fault isolation.
Judges
Diana Montalion, Andrew Harmel-Law, Jacqui Read, Sergey Zinchenko
Team
Sebastian Dąbkowski, Piotr Filipowicz, Wojciech Kasa, Artur Kruszewski
04 / What I teach

Skills I help architects practise.

Most of my teaching happens in workshops, where people work through a realistic problem instead of listening to a lecture. These are the skills the exercises are built around.

  1. 01

    From business goals to architecture requirements

    Turning a business brief into functional requirements and the top three architecture characteristics, with Event Storming as the starting point.

  2. 02

    Finding system boundaries

    Discovering the domain and its bounded contexts before deciding how to split the system.

  3. 03

    Comparing options and justifying trade-offs

    Naming constraints and risks, weighing the options and recording the decision in an ADR.

  4. 04

    Communicating architecture

    C4 diagrams and a short pitch that a business stakeholder can follow and challenge.

  5. 05

    Testing assumptions

    Defending the design in front of a demanding client and answering the hard questions before they come up in a real project.

  6. 06

    Using AI in architecture work

    AI agents that ask hard questions and draft documents from the team's notes, while the people keep the decisions.

05 / In their words

What clients & teams say.

Fourteen written recommendations, from developers who reported to me to clients on the other side of the table. Grouped by what they keep coming back to.

What sets him apart is his ability to bridge the gap between technology and business.

Karol Wiszowatyco-CEO, Spyrosoft Innovo · author of The One Minute Architect
Manager

He was able to fully understand where all of us were coming from, and at the same time offer new, futuristic, way-ahead-of-us kind of solutions.

Vlatka BencinaProject Manager, DPDgroup CSEE
Client
07 / Selected work

Decisions, on the record.

Every project starts with a business problem, not a tech stack. So that's how I document them: as Architecture Decision Records with context, decision and consequences.

ADR-0182025 →Status: Ongoing

Decide before you build

Arpeggia Studio clients · advisor

Context
Companies about to sign a vendor contract, rescue a troubled project or replace a legacy system, with no independent view of the risk.
Decision
Structured discovery and architecture reviews: options with cost, risk and timeline impact; a risk map; an action plan teams and vendors can execute.
Consequence
Decisions made while they're still cheap to change.
My role
Co-founder, senior solution architect
ADR-0142021–2026Status: Live

Contract Lifecycle Management

INNIO · via Inspeerity · architect

Context
Contracts across the company were configured inconsistently and their real costs were hard to predict.
Decision
A CLM platform that configures contracts correctly and simulates their costs before signing.
Consequence
One controlled, unified way to contract, with predictable numbers.
My role
Engineering Team Lead & Lead Architect
ADR-0092015–2018Status: Delivered

B2C platform for 14 countries

DPD Group · via eo Networks · architect

Context
A B2B parcel carrier needed a direct relationship with parcel recipients across 14 countries and 13 business units with different rules.
Decision
Three portals (customer hub, parcel redirection to Parcel Shops/PUDO with tracking, returns) on Spring Boot microservices, configurable per business unit.
Consequence
One platform, local behaviour per country.
My role
Java Architect & Team Lead: architecture, infra sizing, team
ADR-0082014–2015Status: Delivered

One portal for every labour office in Poland

Ministry of Labour & Social Policy · via eo Networks · architect

Context
Hundreds of county and regional labour offices ran separate, ageing portals.
Decision
Replace them with a single nationwide Liferay platform with shared search and workflows.
Consequence
A unified public service across the country.
My role
Liferay Architect & Team Lead

Facing a decision that's expensive to undo?

A new product, a troubled project, a legacy system or a vendor proposal on your desk. Let's talk it through for 30 minutes before you commit budget.

Book a free consultation →piotr.filipowicz@arpeggia.studio