CV
PhD candidate in Computational Linguistics (thesis submitted, awaiting defense). I post-train and evaluate transformer-based speech and language models using SFT, RL, and DPO, and I design the evaluation methodology that validates whether those changes actually improve behavior, not just metrics.
Research interests
Post-training and evaluation of generative language and speech models (SFT, DPO, RL); multilingual representation learning and cross-lingual generalization; evaluation methodology for generative systems (LLM-as-judge, pairwise preference, ablation-based diagnosis).
Work experience
AI Research Engineer, Diabolocom
June 2026–present
- Benchmarked open multilingual TTS systems (Qwen3-TTS, MOSS-TTS, Higgs Audio) for call-center synthesis, focusing on dialogue-turn consistency, voice cloning, and inference-time pronunciation correction for named entities.
- Extending language coverage via supervised fine-tuning, followed by RL-based post-training.
Applied Scientist Intern, Amazon (Alexa TTS)
July 2025–January 2026
- Post-trained a production multilingual TTS model via SFT followed by RL, and designed the evaluation framework used to validate the resulting behavioral changes.
- Built benchmarking datasets and automated diagnostics adopted by research and engineering teams for iterative model improvement.
Research Assistant, Spoken Language Systems (LSV), Universität des Saarlandes
2020–2022
- Ran large-scale probing experiments on multilingual transformer encoders, quantifying how training regime and language typology shape cross-lingual representation transfer.
Research Assistant, Information and Service Systems, Universität des Saarlandes
2019–2020
- Developed time-series neural models for sales forecasting and built the surrounding training and inference pipeline.
Education
PhD in Computational Linguistics, Universität des Saarlandes
SFB 1102, project C4 · 2022–late 2026
Focus: evaluation of neural architectures, representation learning, multilingual modeling, and transformer interpretability.
Advisors: Prof. Dr. Tania Avgustinova, Prof. Dr. Dietrich Klakow, Prof. Dr. Bernd Möbius.
Thesis: Cross-Linguistic Analysis of Microsyntactic Units in Slavic Languages: Computational and Empirical Perspectives (submitted, awaiting defense).
MSc in Language Science and Technology, Universität des Saarlandes
Department of Language Science and Technology · 2019–2022
Advisors: Dr. Badr M. Abdullah, Prof. Dr. Dietrich Klakow.
Thesis: Modeling Mutual Spoken Word Intelligibility between Slavic languages.
BA with Honours in Linguistics, Udmurt State University
Department of Translation and Applied Linguistics · 2014–2018
Technical skills
Post-training and evaluation: SFT, DPO, LoRA and full fine-tuning; RL-based post-training; multilingual adaptation; pairwise preference evaluation with confidence intervals and paired comparison; ablation and error analysis; probing of internal representations.
Frameworks and infrastructure: PyTorch, Hugging Face Transformers, scikit-learn, pandas, NumPy, spaCy, NLTK, langchain; experiment management and reproducible pipelines; AWS (EC2, SageMaker, S3); Docker; Kubernetes; Git; Linux.
Programming languages: Python (advanced), C++ (intermediate), R (intermediate), Bash.
Languages: English (C1), German (B2), Russian (native).
Publications
2025
- It's Not a Walk in the Park! Challenges of Idiom Translation in Speech-to-text Systems ACL 2025 Paper BibTeX Code: idiom_s2t
- Attention on Multiword Expressions: A Multilingual Study of BERT-based Models with Regard to Idiomaticity and Microsyntax Findings of NAACL 2025 Paper BibTeX Code: mwe-attention
- The Effect of Word Predictability on Spoken Cross-Language Intelligibility Interspeech 2025 Paper
- Predictability of Microsyntactic Units across Slavic Languages: A translation-based Study NoDaLiDa Baltic-HLT 2025 Paper BibTeX
2024
- Cross-Linguistic Processing of Non-Compositional Expressions in Slavic Languages CogAlex 2024 at COLING Paper BibTeX Code
- Cross-Linguistic Intelligibility of Non-Compositional Expressions in Spoken Context Interspeech 2024 Paper PDF Code
2023
- Microsyntactic Unit Detection using Word Embedding Models: Experiments on Slavic Languages RANLP 2023 Paper BibTeX Code
2022
- Mapping Phonology to Semantics: A Computational Model of Cross-Lingual Spoken-Word Recognition VarDial 2022 at COLING Paper BibTeX
2021
- How familiar does that sound? Cross-lingual representational similarity analysis of acoustic word embeddings BlackboxNLP 2021 Paper BibTeX
- Do Acoustic Word Embeddings Capture Phonological Similarity? An Empirical Study Interspeech 2021 arXiv
Funding
- Interspeech 2024 Grant
- Small Grant Scheme for Doctoral Researchers of the MGK, CRC 1102
- Saarland Stipendium 2021
Extracurricular
- Summer School on Linguistic Variability, UMLV 2023
- Summer School on Deep Learning in NLP, RANLP 2023