Запускаем афишу образовательных курсов на «Моём круге». Читать подробности
Обо мне

Main areas of study into University:
• Math (Calculus, Fourier analysis, Number Theory, Computational Mathematics etc),
• Physics (Optics, Mechanics, Quantum Mechanics),
• Mathematical Statistics and Data analysis,
• Programming (C/Fortran coding, Data Analysis with R, General Computer Science)
• Astronomy specific cources.

<<Space-kinematic modeling of a homogeneous flat subsystem of the Milky Way>>
In this work I give distance to the center of the Galaxy (Milky Way) by analyzing star catalog data APOGEE-RC. For this I needed to write an extensive program package. Languages such as C and Python were used.

Computer Science Center is an additional education with so relevant courses:
• Data Structures and Algorithm
• C++ Programming
• Concrete Mathematics
• Java Programming
• Python Programming
• Concurrency programming
• Operating System
• GPU calculations

I am rapid adoption of new technologies, with nice git practice, have analytical thinking.

Участие в профессиональных сообществах
Опыт работы
Linux Kernel Developer
Январь 2018 — Сентябрь 2019 (1 год и 9 месяцев)

Optimization of the Linux kernel network stack, implementation of L2 / L3 protocols, creation of additional interfaces for userspace applications, debugging and ensuring portability.


• Optimizations for MPLS. Implementation of the MPLS protocol suffered flaws in

data bases. I have profiled and eliminated bad asymptotics in the data structures

used. This increased the performance of the MPLS stack in 50% to 200% in the

used traffic schemes.

• Fast VLAN change. The basic implementation of VLAN does not allow changing

VLANID and VLAN PROTO without re-creating the child nodes. I implemented

the correct change of the VLAN parameters of the interface without the need to

recreate the child nodes.

• Neighbours caching. In order to increase performance, the neighbors cache was

implemented, which allows to increase performance by about 15%.

• DSA driver for Marvell. I had the experience of writing a DSA driver for a

Marvell switch for a 3.10 kernel, since there was no support for this device yet,

and the reverse porting was impossible due to a serious difference in architecture.

• Lockless into network sched. When identifying bottlenecks, a lot of sequential

code was detected, caused by multiple locks, some of which could be either re-

placed with lighter ones or removed with a modification of the scheduling algo-

rithm. These changes increase productivity quite seriously, ranging from 150%,

but require careful analysis. This experimental has not yet been released.

Technology/Tools: C, gdb, perf, gcc, concurrency programming, algorithms.

Computer Science Center
Сентябрь 2017 — Июнь 2018 (10 месяцев)

Lab of mobile robots provided an opportunity to practice.


• Optimizations for virtual machines. The lab produced programming courses in

Linux, for which an automatic work verification system was used. I managed to

profile and find bottlenecks in the performance of the testing system and speed

up its work by about 50%.

• Linux Kernel Networking Course. I created several tasks for a Linux kernel

programming course that were related to the network stack and its use. Topics

are network interfaces, their use, management, sending and receiving of traffic,

the main data structures of the network stack.

Technology/Tools: C, Bash, Python, valgrind, gdb, perf.

Высшее образование
Санкт-Петербургский государственный университет (бывший ЛГУ)
Факультет: Математико-механический
Сентябрь 2014—Июнь 2019 (4 года и 9 месяцев)


  • Участие в конференции "Современная звездная астрономия - 2017".
Дополнительное образование
Совместная инициатива Computer Science клуба при ПОМИ РАН, компании JetBrains и Школы анализа данных Яндекса