Co-Founder at DeepAvatar, Kaggle Competitions Grandmaster
AI is analyzing your overall score…
Identifying your key strengths…
Evaluating your skill match against the job requirements…
Assessing your cultural and operational fit
I am an engineer and researcher with deep knowledge of modern computer vision. I specialize in building high-quality AI systems that efficiently process large amounts of video data. I've got the Kaggle Grandmaster title for top places in international ML competitions. My experience includes developing the core of an AI system that processes 100,000 live video streams. Kaggle link: https://www.kaggle.com/davletag
Moscow Institute of Physics and Technology (State University) (MIPT)
Master’s Degree, Applied Mathematics and Informatics
January 1, 2015 – January 1, 2017
Moscow Institute of Physics and Technology (State University) (MIPT)
Bachelor’s Degree, Applied Mathematics and Physics
January 1, 2011 – January 1, 2015
DeepAvatar
AI Engineer, Co-Founder
September 1, 2022 – Present
Groningen, Netherlands
NTechLab
Senior Deep Learning Engineer
January 1, 2018 – August 1, 2021
Moscow, Russia
NTechLab
Deep Learning Engineer
September 1, 2016 – January 1, 2018
Moscow, Russia
ABBYY
Software Engineer
September 1, 2014 – September 1, 2016
Moscow, Russia
4th out of 459 teams: NFL 1st and Future - Impact Detection
December 1, 2020 – January 1, 2021
In this competition, I developed a computer vision model that automatically detects helmet impacts in videos of NFL plays. After looking at the data carefully, I realized that in order to achieve a high score, it is necessary to use a temporal context. Therefore, I chose a two-stage approach: helmet detection and ROI classification using a 3D convolutional network.
3rd out of 2265 teams: Facebook - Deepfake Detection Challenge
December 1, 2019 – March 1, 2020
Took 3rd place in the Facebook's Deepfake Detection Challenge using an ensemble of EfficientNet-B7 models. The task of the competition was to detect deepfakes in videos. The main difficulty was that the private test dataset was hidden from the participants and was from an unknown domain, different from the training set. Therefore it was necessary to make the most robust models. I used the mixup technique and heavy augmentations in my solution to make the models generalize well to unseen distributions.
4th out of 468 teams: Google AI - Inclusive Images Challenge
September 1, 2018 – November 1, 2018
The task of this competition was to create an image recognition system that can perform well on test images drawn from different geographic distributions than the ones they were trained on.
Cultural Fit Analysis
The candidate's background is heavily focused on AI/ML research and development, particularly in computer vision. While this demonstrates strong technical depth, the target role is 'Backend Engineer'. There is a potential mismatch in direct backend engineering experience, as the resume primarily showcases AI/ML engineering roles. The project diversity is within the AI/ML domain, which might limit breadth for a general backend role. However, their experience in deploying AI systems implies some exposure to backend infrastructure.
Soft Skills & Operational Fit
The candidate's project descriptions and professional experience highlight a results-oriented individual with a strong drive for innovation and problem-solving. Their co-founder role at DeepAvatar suggests entrepreneurial spirit and ability to take initiative. The descriptions imply strong analytical and technical communication skills, though no direct communication assessment data is available.