Veronika Levdik

- Developed a solution for a sequence semantic multiclass labeling task based on the Transfer Learning of the NLP pre-trained model. - Implemented Python package for Preprocessing, Fine-Tuning and Inference of the LayoutLM-based network. Evaluated AP 93%. - Building training datasets, tuning hyperparameters, gathering statistics, applying postprocessing. - Researching scientific papers (ML, NLP, Graph CNN, Hierarchy Extraction). Embedding Python modules into C++ project.

Tasks like: - Developed an executable application with interactive 3D-visualisation of the 6-axis robot - Implemented a Python package for imitation of a floating dock station motion by modeling water level dynamics in ballast cells. - Developed an optimized algorithm for composing fuel elements in a nuclear power reactor.

Time-Delay Systems, Lyapunov Functionals, Technology of Web Application Development

Linear Algebra, Differential Equations, Mathematical Analysis, Methods of Optimisation, Probability Theoty and Mathematical Statistics, Control Theory, etc