cv

R&D Leader, Executive Director of Data Science

Sber AI, AI-Generated Building Construction

Leading research and development in Generative AI for architecture.

• Developing multimodal models for automated floor plan generation
• Interior layout synthesis and 3D massing generation

Senior Research Scientist, Group Leader

Lomonosov MSU Institute for Artificial Intelligence

Leading a team of 25 scientists in robustness and quality of AI-based image/video processing.

• Developed an attack on the VMAF quality metric, integrated into Google's libaom (AV1)
• Released two prominent benchmarks (#1 on Papers With Code) on video quality measurement

Research Scientist, Group Leader

ISP RAS Research Center for Trusted Artificial Intelligence

Established industry collaboration for deepfake and AI-generated image detection.

• Collaborated with mathematicians for projects in optimization and federated learning

PhD

Lomonosov Moscow State University

Computational Mathematics and Cybernetics

Master

Lomonosov Moscow State University

Computational Mathematics and Cybernetics

Bachelor

Lomonosov Moscow State University

Computational Mathematics and Cybernetics

Winner of the national award for top-3 young scientists in AI in Russia 'Leaders of AI'

Allience for Artificial Intelligence --- leading association of major Russian technology and industrial companies (Sber (Sberbank), Yandex, VK (formerly Mail.ru Group), MTS, Gazprom Neft, and the Russian Direct Investment Fund (RDIF) etc.)

Awarded $13,000.

Lomonosov Moscow State University's stipend for talented young research scientists

MSU

Stipend for talented postdocs and young research scientists.

Finalist of the national award for young scientists 'Leaders of AI'

Allience for Artificial Intelligence --- leading association of major Russian technology and industrial companies (Sber (Sberbank), Yandex, VK (formerly Mail.ru Group), MTS, Gazprom Neft, and the Russian Direct Investment Fund (RDIF) etc.)

Best Presentation Award

Artificial Intelligence and Cloud Computing Conference

Winner of the competition among young scientists in AI, cognitive systems, and brain

Lomonosov MSU and Intellect Foundation

Awarded $30,000.

Winner of the competition 'Start-Artificial Intelligence'

Foundation for Assistance to Small Innovative Enterprises in Science and Technology (FASIE)

Awarded $65,000.

Best IT Project 'Eureka!'

Innopraktika Foundation

Awarded $33,000.

Winner of 'Umnik' competition

Foundation for Assistance to Small Innovative Enterprises in Science and Technology (FASIE)

Conducting research on perceptual 3D video quality metric ($8,000).

From Pixels to Reality: Physical-Digital Patch Attacks on Real-World Camera

PerCom Demo

V. Leonenkova, E. Shumitskaya, D. Vatolin, A. Antsiferova.

Stable VMAF: investigating VMAF's vulnerabilities to adversarial attacks

Multimedia Systems

S. Lavrushkin, M. Khrebtov, A. Antsiferova, G. Bychkov, A. Soloviev, D. Vatolin.

Guardians of Image Quality: Benchmarking Defenses Against Adversarial Attacks on Image Quality Metrics

ICML

A. Gushchin, K. Abud, G. Bychkov, E. Shumitskaya, A. Chistyakova, S. Lavrushkin, B. Rasheed, K. Malyshev, D. Vatolin, A. Antsiferova.

Stochastic BIQA: Median randomized smoothing for certified blind image quality assessment

Computer Vision and Image Understanding

E. Shumitskaya, M. Pautov, D. Vatolin, A. Antsiferova.

Robustness as Architecture: Designing IQA Models to Withstand Adversarial Perturbations

ACM Multimedia Brave New Ideas

I. Meleshin, A. Chistyakova, A. Antsiferova, D. Vatolin.

LEHA-CVQAD: Dataset To Enable Generalized Video Quality Assessment of Compression Artifacts

ACM Multimedia Datasets

A. Gushchin, M. Smirnov, D. Vatolin, A. Antsiferova.

WIBE: Watermarks for generated Images Benchmarking & Evaluation

ASE Demo

A. Yakushev, ..., A. Antsiferova, K. Lukianov, Y. Markin.

Comparing the robustness of modern no-reference image- and video-quality metrics to adversarial attacks

AAAI

A. Antsiferova et al.

IOI: Invisible One-Iteration Adversarial Attack on No-Reference Image- and Video-Quality Metrics

ICML

E. Shumitskaya, A. Antsiferova, D. Vatolin.

Fast Adversarial CNN-based Perturbation Attack of No-Reference Image Quality Metrics

Tiny papers @ ICLR

E Shumitskaya, A. Antsiferova, D. Vatolin.

Video compression dataset and benchmark of learning-based video-quality metrics

NeurIPS

A. Antsiferova et al.

AI-generated building construction

Leading the development of multimodal models for automated architectural design, including floor plan generation, interior layout synthesis, and 3D massing. The project focuses on integrating AI with BIM (Building Information Modeling) to optimize real-time design and documentation analysis.

• Developing generative models for automated floor plans and 3D building massing
• Mentoring researchers on problem formulation and experiment design
• Coordinating cross-functional teams to deliver production-ready R&D outcomes

AI-generated image detection

Researching deepfake and AI-generated image detection methods. Developed a methodology for testing, collected a large-scale dataset with crowdsourced markup, and created a benchmark for evaluating the robustness of detection methods against adversarial attacks.

• Head of the most popular CVPR NTIRE challenge 2026
• Benchmarking robustness of state-of-the-art deepfake detection methods
• Developing new robust detection algorithms

Adversarially robust image/video quality assessment

Created benchmarks for evaluating the robustness of image and video quality metrics against black-box and white-box adversarial attacks. Identified vulnerabilities in 15 no-reference metrics and established a new standard for metric certification.

• Found vulnerabilities in current SOTA full-reference methods
• Developing defense methods including adversarial training and purification
• Published comprehensive robustness benchmarks for the research community

Video quality metrics benchmark for compressed video

Developed the industry's largest benchmark for video-compression-related quality metrics. The dataset includes 2,500+ compressed streams and 780,000+ subjective responses, used by leaders like Google (YouTube), Huawei, and Tencent.

• Released two prominent benchmarks ranked #1 on Papers With Code
• Averages 15 downloads per month from leading industry and research institutes
• Standardized methodology for objective metric evaluation in video encoding

Recognition-aware video quality metrics (with Huawei)

Consulted on the research and development of a novel metric designed to predict object-detection accuracy for compressed videos, shifting the focus from human perception to machine-task performance.