Research · AI · Education

Building trustworthy knowledge systems.

I develop and evaluate local AI systems for research, higher education and knowledge-intensive applications — with a focus on Retrieval-Augmented Generation, task-specific fine-tuning, reproducibility and measurable quality.

Current focus

Research and development

From document collections and retrieval pipelines to specialized language models and transparent evaluation.

RAG

Local knowledge systems

Privacy-conscious assistants that retrieve evidence from an organization’s own documents and make sources visible.

Explore local RAG →
Models

Task-specific fine-tuning

Improving small and medium open-weight language models for grounded answers, hard negatives and structured output.

Explore TSFT-RAG →
Quality

Evaluation and benchmarks

Comparing base and tuned models with realistic test sets, citation checks and reproducible metrics.

Explore evaluation →
Open work

Models, code and documentation

Public model releases, repositories and technical documentation make experiments inspectable and reusable.

Selected work

Projects

Applied research connecting technical development, university practice and open knowledge.

RAG in Teaching

Local AI-supported FAQ systems for teaching, student services and university administration.

Project details →

RAG Memory Bot

An experimental memory and retrieval assistant that connects conversational interaction with controlled document retrieval.

Project details →

LaTeX Resources

More than two decades of courses, examples, package documentation and material for scientific writing.

Browse LaTeX resources →