Cinemersive Labs enables the next-generation of fully immersive, three-dimensional photography and video for VR & AR in six degrees-of-freedom. Throughout my work at CL, I completely dedicate myself to creation of 3D representation of immersive VR videos.

• Self-supervised fine-tuning of MVS network (via reprojections, pseudo GT, synthetic data);• Surface estimation and simplification of a 3D scene represented by point cloud.

The company is engaged in outsourcing ML & CV solutions. As technical director, I was responsible for the following aspects of the company:• Interaction with business customers to form a technical solution, as well as risk assessment during its implementation;• Interaction with several technical teams of customers to build system integration;• Designing a system infrastructure that takes into account the requirements for safety, efficiency and fault tolerance;• Leading the infrastructure and ML departments.

The major research direction during my two years at Skoltech was Part-based Scene Understanding in the 3D Computer Vision field:• CAD-Deform: Deformable Fitting of CAD Models to 3D Scans [published at ECCV 2020, Core A, first shared author]• Towards Part-Based Understanding of RGB-D Scans [published at CVPR 2021, Core A*, second author];• Scan2Part: Fine-grained and Hierarchical Part-level Understanding of Real-World 3D Scans [published at VISAPP 2022, Core B, second author].As part of the research grant, we have developed a monitoring system for a distributed network of RWIS (Road Weather Information System) sensors:• the system predicts the temperature of the road surface with an accuracy of 0.6 degrees for the next hour, which allows the road quality service to promptly decrease the probability of roadway icing• an article was published (ISNN 2018, Springer, Scopus) and 6 IP objects were formed: three patents (RU191374, RU188794 and RU187992) and three computer programs (RU2018662437, RU2018662436 and RU2018662208)

NLP-based analysis of scientific paper and prediction of acceptance to a top conference. • Responsibilities: Product & Customer Development, Lead Researcher;• Winner of The Catalyst Grant, Digital Science, UK, 2019.

For my two years as a junior research scientist, I was dedicated to developing a model-free anomaly detection method which:• provides probabilistic abnormality scores based on the conformal prediction paradigm;• adapts to non-stationarity in univariate data streams (e.g. AWS server metrics, online advertisement clicking rates, Twitter mentions).

• Developing programs for automatically text files generation (C ++) • Generation of mathematical figures based on description

• Developing scripts for automated work with databases• Testing and acceleration of existing methods of working with databases • The thread-safe database integration (Oracle DB, MS SQL)