I am pursuing a PhD in computational linguistics at the University of Tübingen. My primary interests include weak constraints in grammar, abstraction/generalization capabilities of LLMs, and pragmatic reasoning with LLMs. In particular, I am currently investigating whether LLMs can abstract the interaction of conflicting constraints that govern irregular phenomena in German as humans do.
Click here to access my resume.
| 2025-2029 | Doctoral Researcher, Eberhards Karls Universität Tübingen, Tübingen, on-site • Currently occupied at a long-term project called The Status of Linguistic Constraints in Neural Language Models as a part of the LaSTing Priority Program (funded by the DFG, reference number HI 495/8-1). The current step of the project is investigating whether the LLMs are able to acquire weak (defeasible) grammar constraints on the example of German compounding. See DieKonstraints on GitHub. • Instructed the course "LLM-based Assistants" for BA students. See the course page and the repo on GitHub. |
| 2024-2025 | Working Student for LLMs, IPH gGmbH, Hannover, remote • Prototyped custom RAG pipelines (Python, n8n, LangChain, streamlit, docker). |
| 2023-2024 | Working Student for LLMs, IAV GmbH, Berlin, hybrid • Implemented and integrated an internal tool for evaluation of LLMs and RAG pipelines (Python, LangChain, Ragas, OpenAI). • Connected OpenAI tool calling to our internal on-premise models (Python, TGI, LangChain). |
| 2022-2024 | Tutor, Eberhards Karls Universität Tübingen, Tübingen, on-site • Held tutorials and contributed into evaluating assignments and projects for the courses "String Algorithms" (WiSe22/23), "Python for Beginners" (WiSe22/23), "Statistical Language Processing II" (SoSe23), "Reinforcement Learning for LLMs" (SoSe24). • Created Finite-States Transducers for measuring distance between German dialects for the course "String Algorithms" (WiSe22/23) (Python, pynini). |
| 2021-2023 | Data Scientist / NLP Engineer, MKSKOM, Moscow, hybrid • Implemented the backend for a custom RAG engine for a large customer (Python, LangChain, Llama.cpp, HF Transformers). • Increased the number of responses from potential customers on a freelance platform by 5 times by designing and implementing an LLM-based tool for automatic search for and contacting leads (Python, LangChain, OpenAI, Puzzle [internal tool]). Сommunicated with potential customers on that platform. • Implemented prototypes of various tools for NLP, time series prediction, optimization modeling, data analysis (Python, PyTorch, spaCy, scikit-learn, statsmodels, Pandas, NumPy, streamlit). |
| 2026 | Constraints on Linking Element Choice in German Nominal Compounding: A Large-Scale Corpus Study In Proceedings of the Workshop on Structured Linguistic Data and Evaluation (SLiDE), Palma, DOI: 10.63317/24uhjb79y8ov. |
| 2024 | Derbi: Deutscher regelbasierter Inflektor In Studia Slavica XX, Tallinn, ISBN 2674-2160. |
| 2023 | John Boy Walton at SemEval-2023 Task 5: An Ensemble Approach to Spoiler Classification and Retrieval for Clickbait Spoiling In Proceedings of the 17th International Workshop on Semantic Evaluation (SemEval-2023), Toronto, DOI 10.18653/v1/2023.semeval-1.289. |
| 2025-curr | Eberhard Karls Universität Tübingen, PhD: Computational Linguistics, Tübingen • 2026 Talk and publication at the SLiDE workshop at the LREC26 conference. • 2025 Was admitted to the PhD program directly after completing BA. |
| 2022-2025 | Eberhard Karls Universität Tübingen, BA: Computational Linguistics, Tübingen • 2024 Bachelor thesis on applying LLMs to German compound splitting (Python, LangChain, PyTorch, HF transformers). See DEKOR on GitHub. • 2023 Participated at the SemEval 2023 and published a result paper (Python, HF transformers). |
| 2019-2022 | Lomonosov Moscow State University, BA, interrupted: Theoretical and Applied Linguistics, Moscow • 2022 Developed DERBI: a tool for automatic inflection of German words (Python), see DERBI on GitHub. • 2022 Participated at student conferences in Tallinn (on-site), Moscow (remote), Kaliningrad (remote). • 2021 Term paper on predicting ablaut classes of German strong verbs (Python, PyTorch). |
| 2017-2019 | Lomonosov Moscow State University Gymnasium, High School: History and Philology, Moscow • 2019 Prize-winner of All-Russian Olympiad for Linguistics (allows an exam-free admission to a linguistics program at the university of choice). • 2019 Graduation with a Gold Medal. |
| Natural languages | Russian (native), English (C1), German (B2/C1), Spanish (B1), Italian (A2). |
| Programming languages | Python, R, C#, Java. |
| Tech stack | LangChain, HF Transformers, spaCy, PyTorch, scikit-learn, Pandas, NumPy; Docker, GitLab (CI/CD), basic Unix tools. |
| Misc achievements | Prize winner for numerous national and international musical competitions, driving license categories A and B, startup founder (few unsuccessful attempts). |
| Interests | Composing music, traveling, arts & crafts, table tennis, gardening, history, philosophy. |
- I've been living independently since I was 15.
- I gave up Lomonosov MSU to leave for Germany in 2022.
- My best friend and I have made a few unsuccessful attempts to found a startup in 2022-2025.
- In 2019, I became a prize-winner of All-Russian Olympiade for Linguistics, while I had only 5 month to prepare for it.
- The prize gave me a right to enter any university of my choice in Russia for major Linguistics without any exams at all.
- I have completed a musical education, now I compose music from time to time.
- I got my first driving license when I was 17: that one was for motorcycles with 125cc and smaller engines.
- At 18, came back to the same driving school to obtain further licenses: for autos and for large displacement motorcycles; the funny part is, I drove to the exam place on my moto to take an exam for driving motos.
- I learned to weld in 3 days just for fun and crafted a food stand for my dog.
- My mom design patterns for baby knitwear and has 100k+ followers on Instagram (check out BoGatti).