Reference Architecture

A reference stack, adapted to your platform

We start from proven cloud-native patterns, then adapt the stack to your workload, team and operational maturity. GitOps, Kubernetes, operators, secrets, observability and dashboards all live as version-controlled platform logic.

Core delivery pipeline

Source

GitHub

GitOps

Argo CD

Orchestration

Kubernetes

Database

CNPG

Secrets

Sealed Secrets

Observability layer

Collection

Alloy

Metrics

Prometheus

Logs

Loki

Traces

Tempo

Visualisation

Grafana

Developers deploy new services in minutes through self-service scaffolding instead of DevOps tickets. Every component is open-source, CNCF-aligned and deployed through Argo CD. You own all of it from day one.

The Assessment

Reference Architecture

We measure platform debt before we build

Before we encode anything, we identify where engineering time disappears: infrastructure friction, deployment bottlenecks, broken environments, missing ownership, and manual recovery work.

8+h

Per engineer per week

Infrastructure Friction

69% of developers lose 8+ hours a week to broken workflows, manual provisioning and tool sprawl. That is a full workday, every week, spent on work automation should be doing.

Source: State of Developer Experience Report, Atlassian

50+%

Of engineering teams

Deployment Bottleneck

Over half of engineering teams still deploy less than once a week, and when deployments fail, 15% need more than a week to recover. Manual processes and missing automation turn every release into a slow, fragile event.

Source: Google DORA State of DevOps Report 2025

€30k

Per year per engineer

Lost Productivity

Engineers spend 20–30% of their time fighting infrastructure instead of shipping product. Applied to a typical EU SaaS salary, that is roughly €30,000 of lost output per engineer per year.

Source: McKinsey "Why your IT organization should prioritize developer experience"

A €40–80k golden path covers automated pipelines, GitOps, isolated preview environments and self-service workflows. It targets the overhead shown above, and when friction costs every engineer hours each week, even modest reductions pay back fast. The assessment shows where to start.

Talk to an Engineer

In Practice

Three Operating Positions

Platform Patterns We Encode

Preview Environments

One environment per pull request

Every PR provisions an isolated, production-like environment with frontend, backend and infrastructure included. Feature Branch = Environment. The team tests before the merge, and the client sees the feature working exactly as it will in production before it touches the main branch.

release management

Atomic release management

Frontend, backend and infrastructure deploy as a single deterministic unit via GitOps. Every component is versioned and promoted together, so the staging queue disappears along with the question of which version is running where.

Systems Layer

Previews with real third-party servicess

Preview environments that exercise real business scenarios, not only the UI. Our custom Kubernetes operator for Mailgun provisions test domains automatically: DNS records requested from DigitalOcean, Mailgun configured, environment ready. Real email flows, tested without manual setup.

The documentation is the platform itself

Every architectural decision lives in version-controlled logic instead of someone's head or a PDF that goes stale. When we're done, your team can read the system without us.

Start with a focused platform assessment

We map your delivery workflow, infrastructure gaps, platform debt and implementation priorities before anything is built.