Senior product engineering · Munich · Remote EU
I help product teams design, stabilise and ship web, mobile and data systems — from APIs and real-time flows to reliable production releases. One senior engineer, accountable end to end.
TypeScript · Python · PostgreSQL · React Native · AWS — Munich, remote across the EU
Systems in production
Healthcare, industrial IoT, ticketing, retail, marketplaces — each one reached real users.
Interface languages
One care product ships in eleven locales, including Arabic, Turkish, and Ukrainian.
Reused foundations
Auth, config, UI, and utilities written once — the second product shipped without rebuilding the first one’s base.
Every claim on this page is backed by software already running in production.
When this is useful
The date is set, but architecture, data or an integration is in the way, and nobody in the room can say with confidence which one.
You need someone who owns backend and frontend together and makes the call, without adding another permanent headcount to do it.
Real-time flows, APIs and client applications each believe something different about the same record, and support is absorbing the difference.
A larger build is on the table and you want an independent technical read on scope, sequence and risk before committing the budget.
Selected work
Client systems in production, each with the decision that made it work and the trade-off it cost. Where a long-form case study exists, it is linked; where it does not, there is no link rather than a broken one.
2026
ProductionCart, checkout, CMS, and loyalty flows for a production retail app serving Germany and Austria across the mobile and API layers.
Outcome
Market rules stayed in shared APIs rather than being duplicated per country in the app, which made releases safer to trace.
Role
Senior Software Engineer
Kewazo · 2024
ProductionReplaced 30-second polling with live telemetry operators could trust on a construction site — across the ingest, storage, and dashboard layers.
Outcome
Operators acted on the dashboard instead of walking to the machine, and could see when a reading was stale instead of trusting it silently.
2024
ProductionTurned spreadsheet-based clinical analysis into a reproducible web workflow that produced audit-ready reports in minutes instead of a day.
Outcome
Every analysis stored its dataset version, parameters, and pipeline version — so a clinician could regenerate a result and defend it later.
Engagements
Each one ends with something you own outright — a document, a release, a repository your team can maintain. None of them requires you to buy the next.
One system, one critical question, one prioritised plan. You receive a system map, evidence-backed risks, quick wins and a sequenced delivery plan you own.
What you receive
One clearly defined outcome delivered to production — an API or data flow, a mobile feature, a migration, or a reliability improvement.
Senior hands-on capacity inside your product team, with direct ownership of design, implementation, testing and release.
Scope and a fixed price are agreed in writing before anything starts, so the cost of finding out whether this works is known in advance.
How delivery runs
No phase depends on you buying the next one. Each ends with something you own outright — a document, a repository, a release.
Two weeks. We map the domain, name the risks, and produce a written plan with a cost attached. It ends with a document you own — not a dependency on us.
Track Atlas
Schema first, then the surfaces. You get working software in front of real users early, on a foundation that a second product can reuse.
Track Vault · Forge
Before launch, not after the incident. Secrets audited, permissions cut to what you can justify, GDPR posture written down.
Track Sentinel
Stores, pipelines, search visibility, and the handover. Your team gets documentation they can maintain, and the repository keys.
Track Relay · Beacon
Capability tracks
Not six people, and not autonomous software you hire. These are the tracks the work is organised into, so an engagement can be scoped to one of them or run across all six.
Atlas
Maps the domain before anyone writes code, and writes down why each decision went the way it did.
Vault
Designs the schema you will still be able to query in three years, and the access rules that keep it lawful.
Forge
Builds the surfaces users actually touch — web and mobile from one codebase, one design system, one login.
Sentinel
Keeps secrets off the client, permissions minimal, and your GDPR and store paperwork defensible.
Relay
Gets it into the store and keeps it there — pipelines, signing, rollbacks, and the boring migrations nobody budgets for.
Beacon
Makes the product findable — technical SEO, real multilingual support, and analytics you can act on.
Who is accountable
One senior engineer designs the system, writes the parts that carry risk, reviews everything that ships, and signs the release. You talk to that person directly, from the first call to the handover — the same person is on the call, in the pull request and on the incident.
AI and confidentiality
AI tools may accelerate analysis and implementation where the client’s policy permits it. Architecture, security, code review and release decisions remain human-owned. Confidential material is never submitted to third-party AI tools without explicit approval.
Before the first call
Data-intensive web, mobile and backend systems where correctness matters more than novelty: APIs and data models, real-time and IoT flows, healthcare and operational products, retail and marketplace applications. The work is strongest where a system already exists and has to keep running while it changes.
Yes — that is the normal case, not the exception. Embedded Senior Engineering exists for exactly this: working inside your repositories, your review process and your release process, rather than beside them. Greenfield work is welcome too, but nothing here assumes an empty folder.
No. The diagnostic ends with a document, not a retainer. The system map, the risks and the 30/60/90-day plan are yours whether the next phase is bought from Stratos One, given to your own team, or handed to somebody else entirely.
You do. Code is written in your repositories where possible, and handed over with the documentation needed to maintain it without the person who wrote it. No component of a delivery depends on continued access to Stratos One.
AI tools may accelerate analysis and implementation where your policy permits it. Architecture, security, code review and release decisions remain human-owned. Confidential material is never submitted to third-party AI tools without your explicit approval — and if your policy is no AI at all, that is a workable constraint, not a dealbreaker.
From Munich, on Central European Time, remotely across EU timezones. That means a full working-day overlap with teams in DACH and most of Europe, rather than a two-hour window at the edge of the day.
Bring a system diagram, a slow query, a failed release or a product goal. The first 30-minute fit call is free, and if I am not the right person for it, I will say so on that call.
Typical reply time: under 48 hours, from the engineer himself