Iuliia Zaitova – Research Scientist (NLP & Speech)
Open to Research Scientist / Applied Scientist roles in industry.
Saarland University, Germany
PhD, Computational Linguistics (SFB 1102)
Saarbruecken, Germany
Finishing PhD early 2026
Summary
I’m an AI Research Scientist focused on foundation models & transformers for language and speech (LLMs, ASR, TTS). I’m completing a PhD in Computational Linguistics (Saarland University, Germany) and have industry experience as an Applied Scientist intern at Amazon (Alexa TTS, Cambridge, United Kingdom).
Current: foundation models & transformers for language and speech (LLMs, ASR, TTS)
CV · GitHub · Hugging Face
Core expertise
- Production ML systems: large-scale LLMs, ASR, TTS (worked with 10B+ parameter models)
- Evaluation & quality: built frameworks for regression detection and cross-lingual model comparison
- Infrastructure: PyTorch, AWS (SageMaker, EC2), distributed training pipelines
Industry experience
Applied Scientist Intern — Amazon (Alexa TTS, Cambridge, United Kingdom)
- Built and maintained multilingual evaluation workflows to compare models across languages and conditions
- Investigated regression risks and model failure modes to inform deployment decisions
- Developed scalable preprocessing and evaluation pipelines on AWS (SageMaker, EC2, S3)
- Collaborated with applied scientists, engineers, and PMs on production voice systems
Research
- 11 peer-reviewed publications at top ML venues (ACL, NAACL, Interspeech). Research directly applicable to production multimodal AI systems, with focus on model robustness and cross-lingual capabilities.
Currently
- Completing my PhD (early 2026)
- Exploring Applied Scientist / Research Scientist roles in multilingual NLP, speech, and foundation models
- Contact: iuliiazaitova@gmail.com
selected publications
- It’s Not a Walk in the Park! Challenges of Idiom Translation in Speech-to-Text SystemsIn Proceedings of ACL, 2025
- Attention on Multiword Expressions: A Multilingual Study of BERT-based Models with Regard to Idiomaticity and MicrosyntaxIn Proceedings of NAACL, 2025
- Cross-Linguistic Intelligibility of Non-Compositional Expressions in Spoken ContextIn Proceedings of Interspeech, 2024