Project experience
These examples describe my responsibility from architecture and implementation to operation. Operating services, research and planned work are labelled separately. The professional references come from my work at the named employers. Independent developments are presented separately and are not client assignments delivered by RoebTec Solutions.
Embedded Linux migration
Context: M+F Technologies GmbH, embedded product development since 05/2024.
Role: Embedded Systems Engineer.
Challenge: Migrate an existing Windows Embedded Compact 2013 platform to Linux while retaining support for the hardware.
Architecture and implementation: Built a Debian root filesystem, integrated a real-time kernel and ported or optimised device drivers. Automated tests support quality assurance; CI/CD for embedded devices was designed. The role also includes product maintenance and C# development.
Architectural contribution: Shape and implement the transition from Windows Embedded Compact 2013 to embedded Debian, considering existing hardware, driver integration and product maintenance together. The project connects operating-system engineering with the evolution of the platform architecture.
HIL test systems & engineering tools
Context: Molex CVS Bochum GmbH, 04/2019 to 04/2024.
Role: HIL Test Systems Engineer, initially through Excellence AG, directly employed from 11/2020.
Challenge: Develop test systems and quality-assurance processes and support production and maintenance teams.
Architecture and implementation: Designed a test-system concept spanning mechanics, electrical engineering and software. Electrical planning with QElectroTech and database management of test limits and terminal plans. Co-developed C#/.NET test software with MySQL integration; developed a WPF log analyser, PowerShell maintenance tools and Python tests with Modbus integration. Worked in an IATF 16949 / ISO 9001 environment.
Result: Delivered test, analysis and maintenance tools for production and validation. The role also included rework and energy- and cost-conscious solutions.
Federated HIL architecture
Focus: Distributed test execution and integration with industrial equipment.
Contribution: Architecture of a federated test system using runners and PLC integration.
Test management, execution and equipment integration have separate responsibilities. The architecture connects them into a modular system for industrial verification tasks. It demonstrates my work on system boundaries, interfaces and the interaction between software and physical equipment.
This public description stays at the level of architectural principles. Internal implementation details and specific workflows are not disclosed.
ARM/FPGA system integration
Context: Westfälische Hochschule, “Smart Energy Grid Ruhr” research project, 12/2017 to 04/2019.
Role: Research Associate, part-time alongside studies.
Challenge: Integrate embedded system software for a Cyclone V platform combining ARM and FPGA.
Implementation and result: Adapted Debian Linux, developed C/C++ drivers and integrated components into the research system. Provided technical support and advice on embedded systems.
Private Cloud & Distributed Infrastructure
Context: Own decentralized, heterogeneous data centre.
Role: Overall architecture and incremental implementation.
Status: Under construction.
Challenge: Connect virtualization, networking, storage architecture and a service platform into self-operated infrastructure. Different hardware and distributed resources should work together as a coherent system.
Architectural contribution: Define system boundaries and interfaces, account for technical dependencies, and plan operations, access management and recovery together. Technical sovereignty and long-term maintainability guide the design.
Progress: The overall architecture is being implemented incrementally. Services already in operation are described separately in the following section.
Production Infrastructure & Engineering
Context: Own hosting and service infrastructure, operated for over five years overall.
Role: Architecture, implementation, security and operations.
Status: Operating services; the infrastructure’s overall history is not a lifetime or
availability guarantee for every individual service.
Challenge: Operate services under direct control, restrict access and account for maintenance and recovery.
Architecture and implementation: Linux-based DNS, mail and database services, Keycloak as the central identity platform, and Nextcloud and ERPNext business applications. Security zones, segmentation, SSO/OIDC and deliberate exposure restrict service access. Redundancy, failover and backup/recovery are considered for each service individually.
Result: Self-operated service infrastructure with responsibility for integration, updates, access management and operations. RPO/RTO, failover times and availability figures are not claimed without concrete measurements or recovery tests.
Secure Self-hosted Engineering Platform
Context: Own development and documentation environment.
Role: Platform architecture, integration and operations.
Operating today: GitLab and CI/CD with version-controlled MkDocs documentation. Content lives in separate repositories and is built and published automatically. IAM protects internal areas, while approved portfolio and CV content remains accessible without login.
Further development: Connecting requirements, tickets, design and development environments using ERPNext, Nextcloud, Coder and Penpot. This end-to-end process is under evaluation, not presented as a fully deployed product.
Result: The running documentation and build workflow provides a usable foundation; additional integrations are evaluated and introduced separately.
Research / Lab / Evaluation
Private AI & Agent Infrastructure
Local models and GPU infrastructure with LocalAI, MCP tools, Matrix communication and scoped agent identities. The aim is traceable tool access and approvals. Status: Development and evaluation; validation against scoped tasks.
Remote Visualization & Low-Latency Streaming
Own remote-streaming environment using GStreamer, Sunshine and Moonlight. Status: Independent implementation and evaluation; focused on GPU workloads and interactive use.
Ceph & storage architecture
Status: Planned. Design of a six-node Ceph platform, not a production cluster. Data redundancy, failure behaviour and recovery are planning requirements.
Architecture decisions
Decisions are evaluated against requirements, alternatives, consequences and verifiable results. Technical focus areas include:
- KVM/libvirt/QEMU versus an integrated virtualization platform: operating model and integration effort.
- HTTP stack and HTTP/3/QUIC: maintainability, compatibility and operational risk.
- ARM versus x86: workload, performance per watt and lifecycle.
- HA and storage: failure scenarios, RPO/RTO and testable recovery procedures.
These are decision criteria, not published benchmarks or complete architecture decision records.