Hi, I'm
I'm a researcher in Machine Learning and Artificial Intelligence, currently focusing intensively on the development and fine-tuning of large language models (LLMs). My work encompasses the exploration of various AI methods and their practical implementation. With the drive to transform theoretical breakthroughs into practical applications, I am committed to making AI smarter, more reliable, and a little bit less mysterious.
March 2026
Research paper accepted at ICNLP 2026 on systematic evaluation of LLM trustworthiness: measuring output consensus, behavioral consistency, and quality variance, providing methods for explainable, auditable AI systems in regulated environments.
Source: ICNLP 2026
November 2025
Building secure, high-performance LLM infrastructure on local hardware with full data sovereignty: speculative decoding, LoRA batching for multi-tenant scenarios, and transparent agent orchestration for compliance-ready AI at production-grade performance.
November 2025
Interviewed by Northern German Broadcasting (NDR) about the reliability and limitations of Artificial Intelligence in practical applications.
Source: NDR Mecklenburg-Vorpommern
Over the years, I've built complex systems, led teams, and turned vague ideas into working code. Along the way, I've learned to ask the right questions, find the right tools—and occasionally create things that shouldn't have worked in the first place.
How do LLMs work? I explore it through reimplementing GPT architectures, developing evaluation methods to measure consensus and reliability (Quest), and investigating cognitive behaviors in model outputs. Through projects like KiMeKo and DGX Spark deployments, I combine deep technical understanding with practical system building, turning theoretical insights into production-ready AI.
Independent AI systems on local hardware. From high-performance LLM infrastructure on DGX Spark to multi-agent environments where models orchestrate tools and coordinate tasks autonomously. With expertise in PyTorch, Node.js, React/Vue/Angular, C#, and Java, I develop self-contained, performant, production-grade systems for medical AI (KiMeKo) or enterprise monitoring (AnalyticXpert as CTO).
I help organizations implement advanced LLM systems, from multi-agent workflows to evaluating model behavior and reliability. With experience in complex AI concepts and production systems, I bridge the gap between research and practice, specializing in data-sovereign on-premise infrastructure, agentic systems, and deep technical expertise.
Since 2016, I have been teaching cognitive systems, neural networks, and machine learning, combining technical depth with hands-on methods. Developing tools like the FishTank (see projects) has sharpened my skills in knowledge transfer, interdisciplinary communication, and designing scalable, intuitive learning frameworks for complex AI topics.
I have led large-scale projects in both academic and industry settings, managing student teams, supervising PhD candidates, and coordinating complex operations as CTO at Analytic Xpert and on board AIDA ships. These experiences have strengthened my skills in leadership, strategic planning, and cross-functional team management in high-responsibility environments.
I develop creative, research-driven solutions that bridge disciplines and push boundaries, whether through novel teaching tools, experimental AI systems, or custom infrastructures. My work combines curiosity, technical depth, and a strong sense for practical impact.
As a PhD AI researcher with CTO experience, I bridge cutting-edge AI research and practical business implementation. I've deployed 30+ local LLMs in production and trained hundreds of students on AI concepts, which means I can both build robust systems and explain them clearly to your team. I specialize in bringing enterprise-grade AI directly to SMBs without vendor lock-in or cloud dependencies.
I guide and consult companies in setting up large language models (LLMs) on-premise to keep data fully under their control. I lead prototypical implementations and deliver proof of concepts that demonstrate secure, fast, and reliable AI solutions tailored to your specific needs.
I advise and support the development of custom chatbots that leverage your own documents and databases. By creating rapid prototypes and proofs of concept, I help you evaluate how such AI assistants can streamline workflows and improve customer interactions with minimal disruption.
I consult on automating repetitive business tasks using AI agents and oversee the creation of pilot projects. Together, we explore scalable automation opportunities, quickly prototyped and tested, so small and medium-sized enterprises can benefit from smarter workflows and increased efficiency.