Viktor Volkov
Junior Research Assistant
Junior Research Assistant
Junior Research Assistant at AXXX. My current research interests include LLM driven evolution and methods that can help optimize its performance and reduce costs. Also I am interested in Recursive Self-Improvement and novel architectures. Previously I worked on evolutionary neural architecture search for computer vision tasks.
I'm always open to exciting collaborations — feel free to reach out!
My interests:
LLM Agents/MAS Systems
Neural Architecture Search
Evolutionary Algorithms
MSc in CS (honors diploma) - National Research Tomsk State University, Institute of Applied Mathematics and Computer Science (2023-2025)
Master's thesis on "Optimization of Neural Network Architecture Using Genetic Algorithms".
Supervised by Spitsyn Vladimir (link). Thesis results were presented at the following conferences:
12th International Youth Scientific Conference - Mathematical and Applied Software for Information, Technical,
and Economic Systems (link), 2025
International Scientific and Practical Conference of Students, PhD students and Young Scientists
Youth and Modern Information Technologies with award on best research in conference section (link), 2025,
11th International Youth Scientific Conference - Mathematical and Applied Software for Information, Technical,
and Economic Systems(link), 2024
BSc in Physics - National Research Tomsk State University, Faculty of Physics (2018-2023)
Junior Research Assistant, AXXX, Russia, Moscow (2026-06 - now)
Intern, AXXX, Russia, Moscow (2026-01 - 2026-06)
Developed evolution insights extraction module for GigaEvo, an evolutionary framework for automating ML and LLM-oriented tasks.
Developed visualization software for tracking the emergence of successful solutions and the influence of the memory mechanism on evolution.
Conducted evaluation experiments across multiple math and code optimization tasks, including KernelBench (via KernelEvo framework).
Lab Assistant, IMCES SB RAS, Russia, Tomsk (2020-12 - 2022-06 / 2022-12 - 2023-09)
Developed universal software for automated data acquisition and parameter adjustment across multiple spectrometers
Integrated control for CMOS sensor cooling and external laser/sensor units
Participated in assembly, calibration, and testing of an experimental smartphone-based Raman spectrometer
Assisted in experiments with multi-pass optical setups to enhance Raman signal intensity
Developed an algorithm for automated correction of imaging defects in spectrometer data
EvoMem
General purpose memory system for GigaEvo framework
Role: Main contributor
Github/Results: Github (as part of GigaEvo), ArXiv (soon)
"HeroBench" benchmark
12th International Youth Scientific Conference - Mathematical and Applied Software for Information, Technical, and Economic Systems, 2025 (link)
Presentation: "Optimization of Neural Network Architecture Based on the Use of a Genetic Algorithm "(PDF, Russian)
International Scientific and Practical Conference for Students, Postgraduates, and Young Scientists "Youth and Modern Information
Technologies", 2025, (link)
Presentation: "Optimization of Convolutional Neural Network Architecture Using a Genetic Algorithm" (PDF, Russian)
Awarded Best Presentation in Section.
11th International Youth Scientific Conference - Mathematical and Applied Software for Information, Technical, and Economic Systems, 2024 (link)
Presentation: "Architecture Search of a Spiking Neural Network Using a Genetic Algorithm".
22nd International Conference named after A. F. Terpugov "Information Technologies and Mathematical Modelling.", 2023 (link)
Presentation: "Increasing the Accuracy of Raman Spectroscopy Data Acquisition by Using Synthetic Data" Awarded Best Presentation in Section.
No publication.
"Optimization of Convolutional Neural Network Architecture Using a Genetic Algorithm."
V. K. Volkov, V. G. Spitsyn.
Proceedings of International Scientific and Practical Conference for Students, Postgraduates, and Young Scientists "Youth and Modern Information Technologies", 2025
"Raman Gas Analyzer Based on a Multimode Diode Laser." (link)
M. A. Kostenko, I. I. Matrosov, A. R. Zaripov, A. S. Tanichev, V. K. Volkov, S. D. Korkishko, D. V. Petrov.
Journal: Instruments and Experimental Techniques, Pleiades Publishing, vol. 67, pp. 1182-1188, 2024.
DOI: 10.1134/S0020441224702038
"Raman Sensor Based on a Multimode Diode Laser." (link)
M. A. Kostenko, D. V. Petrov, A. R. Zaripov, S. D. Korkishko, V. K. Volkov.
Proceedings of the Conference "Yenisei Photonics – 2024.", p.215, 2024.
"Architecture Search of a Spiking Neural Network Using a Genetic Algorithm." (link)
V. K. Volkov.
Proceedings of the 11th International Youth Scientific Conference "Mathematical and Software Support of Information, Technical and
Economic Systems.", pp.230-236, 2024
Raman Gas Analyzer.", patent (link)
D. V. Petrov, M. A. Kostenko, A. S. Tanichev, V. K. Volkov.
Russian Federation Patent No. RU2787943C1, 2023
Preprints:
P. Anokhin, R. Khalikov, S. Rebrikov, V. Volkov, A. Sorokin and V. Bissonnette. 2025. HeroBench: A benchmark for long-horizon planning and structured reasoning in virtual worlds. arXiv.2508.12782 Retrieved from: https://doi.org/10.48550/arXiv.2508.12782
Summer school "Summer with AIRI", AIRI Institute and National Research Tomsk State University, July 2025
Summer school "Summer with AIRI", AIRI Institute and ITMO University, August 2024
Team's projects “Using LLM to Manage Agents in MMORPG” received best project award.
Virtual Global Winter School, course: "No-Code Artificial Intelligence Tools for Academic Research" Indian Institute of Technology Bombay,
December 2023
Campus course "Introduction to Machine Learning", National Research Tomsk State University, February - May 2021