Keywords:
Generative AI
Large Reasoning Models
LLM Evaluation
Quantization
RAG
Summarization
Human-In-The-Loop
Accessibility
Ed-Tech
LLM Evaluation & Reasoning Validation: Developing methods to assess the reliability of Large Language Models, with a specific focus on validating reasoning traces (e.g., the "LAV" project) and evaluating the capacity of models to act as reliable judges (e.g. project "Do Large Language Models understand how to be judges?").
Computational Argumentation: Analyzing and mining argumentative structures to enhance logic and coherence in AI-generated text, particularly within multi-step reasoning contexts.
AI for Accessibility: Applying Generative AI to remove barriers in scientific publishing, such as the automated generation of descriptions (alt-text) for graphs and complex STEM content.
Retrieval-Augmented Generation (RAG): Integrating external knowledge into generative models to improve factual accuracy and mitigate hallucinations in educational and professional settings.
Writing & Editing Support Systems: Designing writing assistants that adhere to specific editorial standards, covering areas such as Stylistic Adaptation, Abstractive Summarization, and Educational Content Generation (e.g. project "Generation and Evaluation of English Grammar Multiple-Choice Cloze Exercises").
Human-in-the-Loop Workflows: Investigating hybrid Human-AI pipelines to ensure final data quality and trust, especially in high-stakes domains like education and textbook digitization.
