Kai Hauke Krämer

* Lead the design, development, and delivery of scalable AI solutions for financial reconciliation, four-way matching, and accounts payable automation. * Applied NLP & LLM Systems: - Design and implement data-retrieval-augmented, agentic workflows using MCP for document understanding and reconciliation processes - Build and operate production LLM agent systems capable of long-context reasoning, tool use, and multi-step decision-making

* NLP: Developing/training/fine-tuning/maintaining models for the extraction of visually rich documents. In-context-Learning, data-retrieval-augmented agent workflows. * Curation of training- and test-sets. Fine-tune small and large language models (SLMs, LLMs). * Seamless embedding of state-of-the-art AI models in the current project's software stack.

* NLP: Developing/training/fine-tuning/maintaining models for the extraction of visually rich documents. In-context-Learning, data-retrieval-augmented agent workflows. * Curation of training- and test-sets. Fine-tune small and large language models (SLMs, LLMs). * Seamless embedding of state-of-the-art AI models in the current project's software stack.

* Machine Learning Techniques and sophisticated data analysis tools for solving open problems in Dynamical Systems science. * Monte Carlo Tree Search for optimal state space reconstruction * Causal inference, sparse optimization, time series prediction.

* Further development of the Recurrence Analysis framework for time series from complex systems. * Automatized and optimized state space reconstruction from measured time series

Towards a robust framework for Recurrence Analysis: Automated state space reconstruction, optimal parameter selection and correction schemes.