Development and automation
- Python, Bash and shell scripting
- C, C++ and PHP
- Git and CMake
- Catch2, Boost.Test and GoogleTest
Data and integration
- SQL, PostgreSQL and Oracle
- MongoDB
- RabbitMQ, Apache Qpid and ActiveMQ
- Kafka and AMQP
- Imports, exports and validation
Web and e-commerce
- WordPress and WooCommerce
- PHP, semantic HTML and CSS
- Performance, accessibility and technical SEO
- Domains, DNS and SSL/TLS
Systems and operations
- Red Hat Enterprise Linux and Debian
- Docker and Docker Compose
- Kubernetes and MicroK8s
- AWS and Google Cloud
- Server operations, backups and migrations
This describes my technical background, not a blanket support commitment for every product version. Project-specific experience, suitability and capacity are agreed before commissioning work.
Self-hosted AI models and tools
I offer fully self-hosted AI solutions running on company-owned infrastructure. I connect locally hosted models with a customised web interface, internal documents, voice functions and tool integrations. The selection depends on your workflows, existing systems and confidentiality requirements.
- llama.cpp for language-model execution and Open WebUI for chat and knowledge collections
- Custom navigation, a German interface and selectable response profiles
- Local document retrieval with FastEmbed and PostgreSQL/pgvector
- Whisper dictation and Piper speech synthesis on the local server
- Playwright browser tools with manual takeover, scheduled tasks and custom interfaces
- Container resource limits, SSH-tunnel access, documented versions, backups and rollback
- Tests for chat, document questions, speech and tool calls
Model processing, document indexing and speech run without external model APIs. Browser functions can visit external websites. Permitted data flows, access and functions are agreed separately for confidential use. A completely isolated deployment is a separate operating profile; performance and hardware depend on the model and task.