Skip to content

Architecture · Technical leadership · Delivery

Your technical problems, solved end to end — from first meeting to production.

I'm Jakub, a software architect and technical lead with 7 years of experience. I work with your team to understand the problem, design the solution, lead the people building it and deliver it to production — mostly backend systems, integrations and internal tools.

20-minute call · no preparation needed

A modular system: one core, independent modules around itWeb appPartner APICoreProvider AProvider BEventsQA canvas
  • Backed by Fragments s. r. o.
  • Based in Brno, Czech Republic (EU)
  • 7 years as architect & team lead
  • Working with EU & US teams

Sound familiar?

Three bottlenecks that slow engineering down

Most teams I talk to are not short of ideas. They are blocked by the system they already have.

  • 01

    Every integration takes months

    A new feature, third-party API or payment provider means touching a tightly coupled core. Delivery slows down and every change feels risky.

  • 02

    Testing does not keep up

    Testing is spread across services, devices and simulators. QA becomes the release bottleneck — and bugs still reach production.

  • 03

    Backend team, no frontend capacity

    You need an internal tool or web app on top of complex backend data, but hiring a frontend team for it is slow and expensive.

Services

One answer for each bottleneck

Three focused engagements. Every one includes analysis, stakeholder alignment and technical leadership — not only code — and ends with software running in production.

  • Backend architecture & modernisation

    I design modular, event-driven services and replace rigid legacy bottlenecks step by step, so new integrations take days instead of months.

    • Architecture design and review
    • Modular core with isolated integration modules
    • Gradual replacement of legacy parts
    • Integration standards your team — and AI agents — can follow

    Solves 01 · Every integration takes months

  • Engineering tooling & QA platforms

    I build internal platforms — including visual, node-based test builders — that let your team design, run and report on thousands of tests from one place.

    • Visual test builders and node-based canvases
    • Bulk and nightly test runs with central reporting
    • Real-time collaboration for QA teams
    • AI assistants that help write and explain tests

    Solves 02 · Testing does not keep up

  • AI-accelerated full-stack delivery

    I define the data contracts and use AI-assisted development to build the UI layer quickly. You get an interactive web app early, so the architecture is validated with real users, not slides.

    • Data contracts designed backend-first
    • React frontends built with AI-assisted workflows
    • Proven, reusable component library
    • Engineering review before anything ships

    Solves 03 · Backend team, no frontend capacity

Selected work

Proof, not promises

Two systems from the payments domain — the problem, what was built, and what changed.

  • Core service with shared modules and isolated provider modulesGateway APICoreShared modulesProvider AProvider BProvider C
    Payments

    Payment provider integrations: from months to days

    A new modular payment service and AI-ready integration standards cut the time to add a payment provider from months to about a week.

    Time to integrate a new provider
    Months → days
    Analysis, development and QA per provider
    ~5 days
    Development effort per provider
    1 day

    Read case study

  • GreenGuard test flow: a start node, a setup step, a terminal request and a response assertion connected on the canvas.
    Payments · QA automation

    GreenGuard: one visual platform for testing a payment ecosystem

    A collaborative, node-based QA platform that replaced fragmented manual testing across apps, terminals and simulators.

    Complex tests managed in one place
    10,000+
    Bulk runs with central reporting
    Nightly
    Multiplayer editing with presence
    Real-time

    Read case study

How I work

From problem to production

I own the problem, not just the tickets — from the first conversation with the people who have it to a system running in production.

  1. 01

    Understand

    Scoping call + workshops

    I meet the people who have the problem, map the constraints and agree on what success looks like.

  2. 02

    Design

    Weekly check-ins

    Architecture, plan and trade-offs — validated early with a working demo, not slides.

  3. 03

    Build & lead

    Monthly syncs + async

    Hands-on engineering while leading your developers, or trusted specialists I bring in. Progress is visible in your repositories.

  4. 04

    Deliver

    Weekly check-ins

    Production release, handover and documentation, so your team can own and extend the result.

  • Async-first

    Day-to-day communication happens in GitHub or GitLab — issues, pull requests and reviews.

  • Scheduled calls

    10:00 or 14:00 CET for European teams, 17:00 CET (11:00 ET · 08:00 PT) for US teams.

  • On-site when it matters

    Kick-offs and key milestones can happen in person. From Brno, Vienna, Bratislava and Prague are a couple of hours away, and Germany is an easy trip.

How I deliver faster

AI speeds up delivery — judgement stays human

The thinking — problem, architecture, data contracts — is done by people. AI tools and custom agent workflows, paired with a library of proven components, then take over the repetitive build work, so you get results sooner without a lower engineering bar.

  1. 01

    Architecture & contracts

    Boundaries, data models and APIs are designed by hand. This is where quality is decided.

  2. 02

    AI-assisted build

    Agents work from written standards and proven components, so output is consistent and reviewable.

  3. 03

    Review & hardening

    Every change is reviewed and tested like any senior engineer’s work before it ships.

Backend

  • Java
  • Kotlin
  • Spring Boot
  • Node.js
  • Python

Data & infrastructure

  • Kafka
  • PostgreSQL
  • MariaDB
  • Docker
  • Kubernetes

Frontend

  • React
  • TypeScript
  • Tailwind CSS

AI tooling

  • Claude
  • Augment
  • Custom agents
  • 7 years of experience
  • Team lead · Architect · Product owner
  • Brno, Czech Republic (EU)
  • Fragments s. r. o., VAT-registered

About

Jakub Sumihora

Software Architect & Technical Lead · founder of Fragments

I work like a staff engineer: I take a problem, talk to the people who have it, find the right solution and see it through to production. Architecture, trade-offs and the people around the code matter to me as much as the code itself.

I have spent about 7 years building production systems — as a developer, architect and team lead, with a period as product owner. Much of that time has been in payments, where reliability is not optional. I enjoy leading and mentoring teams of up to around ten engineers while staying hands-on.

Fragments is a personal-led studio. You work directly with me — no hand-offs, no account managers — with the contracts, invoicing and accountability of an EU company. I lead your engineers, or bring in trusted specialists I have worked with when the project needs more people.

Fit

Where I add the most value

Being clear about this up front saves both of us time.

A good fit

  • Problems that need someone to own them end to end
  • New products, platforms and modules
  • Integration-heavy systems — payments, third-party APIs
  • Internal tooling and QA platforms
  • Replacing legacy bottlenecks with a new modular architecture
  • Teams of up to ~10 engineers that need technical leadership

Not a fit

  • Staff augmentation — extra hands to close tickets
  • Marketing websites, e-shops or template sites
  • WordPress or PHP maintenance
  • Like-for-like migrations without architectural change
  • 24/7 on-call duties

FAQ

Common questions

How is pricing set?

Per project, on request. After the scoping call I propose milestones with a fixed scope where possible. There is no public rate card because scope, risk and timeline drive the cost.

Who owns the code and IP?

Everything specific to your business is yours. General-purpose tools and components I bring into the project stay mine and are licensed to you for use in it.

How do contracts and invoicing work?

Through Fragments s. r. o., a VAT-registered company in Slovakia (EU). Standard B2B invoicing; EU reverse charge applies where relevant.

Which time zone do you work in?

Central European Time — I work from Brno. Calls are at 10:00 or 14:00 CET for European teams and 17:00 CET (11:00 ET · 08:00 PT) for US teams.

Can you lead or mentor our team?

Yes. I have led teams and can lead or mentor up to around ten engineers alongside hands-on architecture work.

Do you sign NDAs?

Yes — before any confidential details are shared, if you need one.

Next step

Start with a 20-minute scoping call

  1. 01

    Book a call

    Pick a slot that suits your time zone.

  2. 02

    Share a brief

    If there is a fit, you send a short project brief.

  3. 03

    Agree milestones

    We define scope and milestones, then start.