Dr. Daryna
Dementieva
I'm a postdoctoral researcher at the Technical University of Munich, working on Responsible AI and Language Technologies. Now, I mostly focus on more rigorous and trustworthy AI evaluation for good and multilingual, low-resource language technologies. My previous research combats harmful content online — toxicity, hate speech, and misinformation — while advancing the democratisation of language technologies for underrepresented languages, including Ukrainian. I hold a PhD from Skoltech in Computational and Data Science and Engineering and have authored 50+ peer-reviewed papers, including several position papers on NLP for Social Good, massive multilingual benchmarks recognised with the ACL 2025 Best Resource Paper Award and by UNESCO. I am a huge supporter of opensource and various Python and AI communities. I organise the NLP for Positive Impact workshop, the TextDetox and LLMs with Limited Resources shared tasks, and Munich🥨NLP meetups.
Highlighted Projects
NLP4SG: AI and Language Technologies for Good
Survey of nine domains where NLP can serve society, plus the annual NLP for Positive Impact workshop.
Learn more →
TextDetox: Multilingual Text Detoxification
Parallel detoxification corpora, models, and shared tasks — over 2M downloads by the community.
Learn more →
HatePRISM: Proactive Hate Speech Mitigation
Connecting regulations, platform policies, and NLP datasets towards unified mitigation.
Learn more →
UkrDetect: Text Classification for Ukrainian
Datasets and models for toxicity, formality, NLI, and emotion detection in Ukrainian.
Learn more →
Low-Resource NLP
WMT shared task on limited-resource LLMs and an LREC tutorial on inclusive corpora.
Learn more →Currently working on
Rigorous multi-turn evaluation of chatbots for societal domains
Evaluating whether LLMs reason “for good”
Cross-lingual transfer and distillation of reasoning knowledge
Interviews & Videos
How NLP can help healthcare, education, climate, and the fight against disinformation
Read →