Teaching Philosophy

My life shaped my teaching and supervision philosophy. We all start from different places and different backgrounds. It is unfair and unwise to judge people solely on the knowledge — merely facts — that they hold at a given moment. We live in a rapidly changing world. Many facts, knowledge, and skills become outdated not over 50 years, but over 5–10 years, or even within 1–2 years. People who only “know” do not get through these changes. My main teaching outcome for students is to learn how to learn.

Teaching philosophy explained via physics

Let me explain it with physics. One particle (a) moves at a stable speed in a stable direction (uniform motion, a = 0). But will it be able to adapt to new goals, to a new world? Another one (b) is not only constantly moving, but also constantly increasing its speed and even changing direction (accelerated motion, a = C). That one is already more flexible in this developing world. The final one does something even crazier: it is able to change how much it changes its speed (non-linear accelerated motion, a = Ct). This is the effect I want to achieve when working with students — to improve their ability to learn.
Be non-static, always move, improve your movement. Then you can achieve anything and move anywhere you want in this world.

Courses

I have taught several types of courses — large lecture courses as well as practical courses and seminars.

Responsible AI and Language Technologies — Seminar

TU MunichSince SS 2026Main organiser~20 students

A seminar series covering Responsible AI and NLP from many sides: ethics, fair data collection, technical papers on AI for Social Good, and philosophical discussions about designing technology and AI for people.

Projects in Natural Language Processing — Practical Course (BSc & MSc)

TU MunichEvery semester since 2022Main organiser since WS 2025; teaching assistant~50 students

Students in groups of three carry out practical projects on part of a real research project or participate in shared tasks. Several course projects resulted in papers at top-conference workshops ([example 1], [example 2]).

Advanced NLP — Lecture Course

TU Munich2022–2024Lecturer, teaching assistant~200 students

Lecture course on advanced topics in NLP.

Introduction to NLP — Lecture Course

TU Munich2022–2024Lecturer, teaching assistant~200 students

Introductory topics in NLP.

Shared Tasks

I led or was part of the organising team of several multilingual shared tasks:

TextDetox @ CLEF 2025

Task
Toxic → non-toxic style transfer
Languages
15
Teams
30

LLMs with Limited Resources @ WMT 2025

Task
Multi-task tuning of smaller LLMs for a language
Languages
2
Teams
5

BRIGHTER @ SemEval 2025

Task
Detecting basic emotions in text
Languages
28
Teams
150+

TextDetox @ CLEF 2024

Task
Toxic → non-toxic style transfer
Languages
9
Teams
30

Student Supervision

Since the start of my PhD, I have supervised several dozen students from around as many countries, at all levels — Bachelor, Master, and PhD. They have gone on to highly successful careers: PhD positions, lead developers in companies, and start-up founders.

Bachelor thesis Master thesis
StudentTypeTitlePaperNext position
Evgeniya Sukhodolskaya Master thesis
TUM, 2024
A Unified Approach to Cross-lingual News Similarity Evaluation: from Crowdsourcing to Large Language Models Optimization arXiv 2025
Diego Miguel Lozano Master thesis
TUM, 2024
Explainable Textual Similarity Metrics via Equivalent and Erroneous Span Detection LREC 2026
Arda Andirin Master thesis
TUM, 2024
Explainable Fact-Checking in News Datasets with Large Language Models —
Shaghayegh Koli Master thesis
TUM, 2024
Fake News and Machine-generated Text Detection for Persian and English NLP4PI 2024
Valeriia Khylenko Master thesis
TUM, 2024
Cross-lingual Text Classification Knowledge Transfer to the Ukrainian Language COLING 2025, WOAH 2024
Adam Rydelek Master thesis
TUM, 2023
Sexism in Social Media — Explainable Detection Using Deep Learning Models SemEval 2023
Daniel Schroter Master thesis
TUM, 2023
Discovering Human Values Behind Arguments — A Case for Transformer-based Models and Explainable AI SemEval 2023 🏆
Yen-Yu Chang Master thesis
TUM, 2023
Exploration of Approaches to Counter Hate Speech: The Case of Sexist Speech LREC 2026
Yuliana Poliakova Bachelor thesis
TUM, 2023
Detecting the Persuasion Techniques in Online News —
Daniil Moskovskiy Master thesis
Skoltech, 2022
Supervised Text Detoxification Methods ACL 2022 SRW
Mikhail Kuimov Master thesis
Skoltech, 2022
Multilingual and Cross-lingual Approaches to Processing of News Texts SemEval 2022