teaching | Michael Färber's Research Group

Teaching in my group is research-driven and closely connected to our work in:

Scholarly/Scientific LLMs (RAG, citation-aware & verifiable generation)
Knowledge graphs & structured knowledge
Graph machine learning & hybrid neuro-symbolic methods
Trustworthy AI (robustness, evaluation, bias & explainability)

Why join our courses / thesis topics?

Research-first, publication-oriented: real research questions (no toy assignments), often aligned with current lab directions (see /projects/). The default goal is a joint research publication (see /publications/ and /news/); many student projects in our group have already led to papers.
Research-grade artifacts: reproducible code, strong baselines, and rigorous evaluation (incl. ablations / error analysis).
Visibility and outreach: strong outcomes are presented via talks/demos and communicated publicly; when suitable, we also aim for broader outreach.
Close mentoring: supervision slots are limited; we therefore prioritize strong fit and clear commitment.

Student-led papers (selected):

SQuAI: Scientific Question-Answering with Multi-Agent Retrieval-Augmented Generation — CIKM 2025 (demo) · students: Ines Besrour (M.Sc.), Jingbo He (M.Sc.) — Multi-agent RAG framework for scientific QA with inline citations and supporting evidence.
ComplexTempQA: A 100m Dataset for Complex Temporal Question Answering — EMNLP 2025 · students: Raphael Gruber (M.Sc.) — 100M-scale benchmark for complex temporal question answering (Wikipedia/Wikidata).
In-Context Learning for Information Extraction using Fully Synthetic Demonstrations — XLLM@ACL 2025 · students: Ashish Kangen (M.Sc.) — Synthetic demonstration generation + retrieval-based in-context learning for document-level IE.
SimplifyMyText: An LLM-Based System for Inclusive Plain Language Text Simplification — ECIR 2025 · students: Kyuri Im (M.Sc.) — LLM-based system for plain-language simplification with configurable audiences and input formats.

Open artifacts (selected):

SQuAI (code) — Code · students: Ines Besrour (M.Sc.), Jingbo He (M.Sc.) — Multi-agent RAG system for scientific QA with citations.
SQuAI (demo) — Demo · students: Ines Besrour (M.Sc.), Jingbo He (M.Sc.) — Interactive demo of scientific QA with traceable sources.
ComplexTempQA (dataset + code) — Dataset · students: Raphael Gruber (M.Sc.) — Repository for the 100M-scale temporal QA benchmark.
CoDy (code) — Code · students: Daniel Gomm (M.Sc.) — Reference implementation for counterfactual explanations on dynamic graphs.

Note: “M.Sc.” / “B.Sc.” indicate the student level at the time of the work.

More examples: /student-outcomes/.

How we work (projects → papers): /how-we-work/.

Semester Course Offerings at TU Dresden

Each semester page contains the concrete course list and links to course pages:

If you mainly look for the LLM lecture, start here: /secretllm/

Theses & Student Projects

I have independently supervised over 60 theses (Bachelor’s, Master’s, and Diploma).
Current open topics can be found on the student thesis page.
I currently supervise 6 Ph.D. students at TU Dresden.
See the team page for more details.

I am also open to supervising highly motivated students from other German universities (and international students).
Please ensure that any formal requirements (registration rules, examiner requirements, deadlines) are clarified with your home institution.

How to contact us (quick checklist):
Please email the topic’s contact person with:

short CV
transcript of records (and, if applicable: transcript of your bachelor’s)
short statement (a few sentences) highlighting your skills, motivation for the topic, and your planned thesis starting date
if applicable: PDF of your bachelor’s thesis
optional: GitHub/portfolio or a pointer to relevant past projects

If you already have your own topic idea, feel free to send a short (max. 1 page) sketch; we can align it with our research directions.

If you are looking for research assistant / HiWi openings (or other positions), please see /vacancies/.

Administrative Questions

For study-program related questions (e.g., formal requirements, forms, deadlines), this is typically in the responsibility of the student. If needed, please clarify directly with the relevant TU Dresden student examination office (Prüfungsamt) (or the equivalent office at your home university).

For requests to the secretary, please see the contact page.

Past Courses & Teaching Positions

Over the course of my academic career, I have taught a wide range of courses across institutions and disciplines:

KIT — W3 Deputy Full Professor (Oct 2020 – Mar 2023):
Taught B.Sc. and M.Sc. courses in Business Informatics, Business Engineering, and Computer Science — including large-scale courses with up to 600 students.
Postdoctoral teaching (KIT / University of Freiburg / Kyoto University):
Delivered lectures and seminars with class sizes ranging from 150 to 600 students.
Ph.D. teaching experience (KIT, 2012 – 2017):
Led exercises and seminars involving multiple teaching assistants and coordinated small-group supervision.