Student at St. Petersburg State University, Mathematics and Mechanics Faculty, Software Engineering Department.
Areas of interest: deep learning, GPGPU programming.
The aim of the project is to create a tool for code completion. Distinctive features will be the use of the latest architecture of neural networks, such as GPT-2, and the use of a huge amount of data.Project supervisor: Timofey Bryksin
The goal of this project is to improve code synthesis methods. We are currently actively exploring how to apply the latest neural network architectures from the NLP domain and how to efficiently use the tree-like code structure.Project supervisor: Timofey Bryksin
Recently proposed matrix multiplication based algorithm for context-free path querying (CFPQ) offloads the most performance-critical parts onto boolean matrices multiplication. Thus, it is possible to achieve high performance of CFPQ by means of modern parallel hardware and software. In this paper, we provide results of empirical performance comparison of different implementations of this algorithm on both real-world data and synthetic data for the worst cases.Proceedings of the 2nd Joint International Workshop on Graph Data Management Experiences & Systems (GRADES) and Network Data Analytics (NDA), June 2019