I’m a PhD student in University of California, Santa Barbara, working under supervision of Prof. Arpit Gupta in Systems and Networking Lab.
My current research areas include Machine Learning (and Reinforcement Learning specifically) in Networking, including explainable AI, network data representation techniques, reproducibility and scalability of networking active measurement experiments, and ML-based performance inference.
I develop netUnicorn project and PINOT infrastructure to simplify developing credible ML solutions in networking (see PINOT).
I also researched a Machine Learning models credibility in Networking area, exploring vulnerabilities and problems in published existing solutions and how to fix them using Explainable AI methods (see TrusteeML).
- Machine Learning in Networking
- + Reinforcement Learning
- + ML-based performance inference
- + explainability and robustness
- Reinforcement Learning in general
- World Models (including Dreamer) approaches
- Deep Space Network and extra-terrestrial issues in networking
Projects and Publications
For detailed information on publications see: Publications
netUnicorn simplifies (iteratively) curating high-quality data for different learning problems from diverse network environments. It also provides abstractions and infrastructure interaction for developers of data collection pipelines and measurement experiments.
Trustee is a decision tree -based framework for explainable AI, inspired by VIPER project. We showed that many public ML models in networking community are prone to spurious correlations and problems of underspecification, and proposed how to address these issues using XAI techniques. Contribution: public ML models dissection and verification, experiments reproducibility
Tortilla is an Apache Guacamole based project developed to give users access to their virtual machines located on Hyper-V hosts. This project provides a virtual graphical console through RDP/ssh and uses html5 to be available in modern web browsers.
It’s like a VMware web client, but for MS SCVMM.
Contribution: system architecture, system controller (python), scvmm api access (c#), team leading.
Simple client-server app for heavy file hosting and key-based access. Developed for Higher School of Engineering.
Contribution: system architecture, server part (c#, .net core), team leading.
Hackathons & Challenges
- Junction 2018 – runner-up (in-team) in QOCO challenge
(PlanPlanner – redirecting air traffic with graph algorithms)
- World-IT-Planet 2018 – winner (solo) of the international stage in Cloud Systems track
- BioHack 2018 – Special award (in-team) by Institute of Bioinformatics
(predicting 3D cell structure with ML)
- Junction 2017 – winners (in-team) of track “Robots, Learning machines”
(Mirror – applied ML to Pepper robot to make him copy all our movements)
- World-IT-Planet 2017 – winner (solo) of the international stage in Cloud Systems track
(yes, again :D)
- Make your university 20.35 – winners (in-team)
(recommender system for people looking for a job)
- Ph.D. in Computer Science – University of California, Santa Barbara 2021 - now, with Prof. Arpit Gupta
- M.S. in Computer Science – Peter the Great Saint-Petersburg Polytechnical University
(Thesis: applying natural gradient descent to reinforcement learning algorithm)
- B.S. in Computer Science – Peter the Great Saint-Petersburg Polytechnical University
(Thesis: creating music with generative adversarial networks)
- Additional courses:
- Machine learning
- Algorithms and data structures
- Probability theory
- On-Device ML Model Researcher at Huawei Russia (Mar. 2020 - Sep. 2021)
- Junior Reinforcement Learning Researcher at JetBrains Research (Oct. 2019 - Jun. 2020)
- Python tutor at Higher School of Engineering (Sep. 2018 - Sep. 2020)
- System administrator at Higher School of Engineering (Jul. 2015 - Jun. 2019)
- Microsoft Certified: Azure AI Engineer Associate, Microsoft
- Certified Associate at Python programming, OpenEDG
- Microsoft Certified Solutions Expert: Cloud Platform and Infrastructure, Microsoft
- Microsoft Certified Solutions Associate: Windows Server 2016, Microsoft
- Microsoft Certified Professional, Microsoft
- Cisco CCNA Instructor, Cisco
- Cisco Certified Network Associate Routing and Switching, Cisco
- Huawei Certified Network Associate Storage, Huawei
Thanks for your attention
Have any questions? Feel free to contact me in any way convenient for you.