Developing default, fraud and bankruptcy prediction models.

• Developed data pipeline to process text data and match databases using LLM embeddings by BERT, LLAMA2, and machine learning classification models using Python-based algorithms and scikit-learn, pandas, nltk, genism, and TensorFlow libraries with an accuracy of 92%.• Applied OpenAI GPT4 API as an expert data labelling and cleaning system. • Trained Siameze Neural Networks to determine whether similar entries correspond to the same product in the database.• Implemented multiprocessing optimization approach to the pipeline and optimized calculations of embeddings and similarity using GPU with PyTorch.• Assembled a cluster of 24 virtual machines on GCP for fast mining online-based data utilizing Docker containerization.

• Led innovative research and development (R&D) projects in developing digital services for education, such as an AI voice assistant, helping more than 12,000 students find the current class schedule by voice requests to their phones integrated with Yandex voice assistant Alice and utilizing YaLM (Yandex LLM).• Led the development and launch of a service for more than 35,000 university users, which improved the educational success of first-year students by 12% in mathematics. This service utilized large-scale data from a learning management system and machine learning models such as neural networks, KNNs, and Random Forests. • Developed ETL pipeline for data gathering by API from social media platform VK, utilizing Python and PySpark and storing it in MySQL and MongoDB databases on a private cloud.• Based on Power BI and SQL databases, constructed a solution for the student management system that decreased the time of information processing from creation to executive level dashboards to less than 24 hours, and forecasting success model based on academic achievements, time series analysis, SVM, and KNN methods to detect behaviour patterns.• Launched the scientific laboratory for Data Science in Social Research, assembled the team, and prepared the research program based on industry research, best practices, and analysis of scientific publications.• Managed several teams of software engineers, data scientists, analysts and researchers in software development, scientific and educational projects. • Defended digital strategy and data-driven services development projects in the Feredal Ministry, which led to funding.

• Developed the scientific method and software to design cities protection from small coal dust using mathematical physics equations, finite volumes method, and C++ programming with OpenMP and MPI parallelization libraries utilizing distributed supercomputer infrastructure.• Conducted research on using neural networks and genetic algorithms to optimize the shapes of aerial vehicles, prepared scientific articles, and launched a student fabrication laboratory for modelling, programming, and prototyping unmanned aerial vehicles.

• Led a team of 12 engineers to launch a Platform as a Service for over 5,000 supercomputer users, including academic researchers worldwide and computer and data scientists, to provide high-quality customer service, according to ITIL. • Launched a private cloud based on Panasas, IBM and DELL solutions for the organization’s data engineering purposes and data storage reliability.• Supervised and mentored more than 30 students during their research projects and internships in network engineering, software engineering, data science, machine learning, and infrastructure automation software.

• Developed the main module using C++ for pollution prediction in coal mines modelling software, connecting other modules based on different stacks, such as Fortran, C++, and Delphi.• Configured and maintained Linux-based server infrastructure for a high-performance cluster of more than 500 nodes.• Configured active network hardware Dell, Dlink, Juniper, and Infiniband for HPC cluster with more than 500 nodes.• Maintained the electrical and cooling infrastructure of the supercomputer based on APC, MGE and Stulz hardware

• Produced the software to model and visualization of self-propagating high-temperature synthesis using Delphi.