Bio.
My name is Nikita, I am a Co-Founder and CTO at OHMYSYNT and a PhD student at ITMO University.

Since the very beginning of my career, I've been doing everything from the ground up: designing prototypes, managing a development team, and decomposing upcoming goals. For the last 4 years, from the moment I left the corporate world for the startup world, I've been solving only cutting-edge AI problems, earning my Master's and PhD at the same time. As a result, optimization became my life's credo: I always had to utilize minimal resources and achieve maximum impact. Therefore building things efficiently is a particular strength of mine.

Now I'm a Founder with 500 Global Accelerator experience, sharing ideas at tech conferences, learning how to grow a business and trying to create something people really need.

Since 2021, my responsibilities as CTO included: While obtaining the PhD, I focus on the Generative AI.
While completing the Master's program, my field of research included infrared aerial images segmentation, explainable AI in medicine and optimization of the training process of the Vision Transformer.

Work experience.
Jan 2023 — Present : Co-Founder and CTO at OHMYSYNT - Creating a personalized AI content generation service
Feb 2021 — Dec 2022: CTO at Malivar - Creating service with synthetic avatars for businesses and content creators
Jan 2022 — Jul 2022: ML Researcher at ITMO University - Researching explainable AI in medicine
Jan 2021 — Jul 2021: ML Engineer at ITMO University - Implementing segmentation model for drones
Mar 2020 — Feb 2021: ML Research Engineer at Malivar - Creating virtual humans with Deep Learning
Sep 2018 — Apr 2020: Data Scientist at SberBank - Optimizing employee productivity
Jul 2018 — Sep 2018: Data Science Intern at SberBank - Satisfying business needs with Data Mining

Education.
2023 — 2023: Founder at 500 Global Accelerator Program - Startup Growth
2022 — 2025: PhD at ITMO University - Artificial Intelligence and Machine Learning
2020 — 2022: MS at ITMO University - Applied Informatics in Big Data and Machine Learning
2016 — 2020: BS at Ural Federal University - Applied Informatics

Conferences and Talks

2023 DataFest Yerevan Conference — Text-to-Video Generation and Editing. Video.

2023 ODS DataFest Conference — An overview of the timeline of the Text-to-Video Generation and Editing. Video.

2023 AINL Conference — Short paper: Improving Text-to-Image Models Personalization on People with Face-specific Objectives. Video.

2023 ML Lab ITMO Seminar — Fine-tuning of Text-to-Image Models: DreamBooth and Textual Inversion. Video.

2019 Deep Learning Seminar — Zero-Shot Learning. Slides. Video.

2019 Deep Learning Seminar — ESPCN and approaches for the Super-Resolution problem. Video.

2019 Deep Learning Seminar — Gated Shape CNN for Semantic Segmentation. Video.

Publications

Exploring the Relationship Between Error and Interpretation of the Segmentation Model's Prediction

Nikita Detkov, Ksenia Balabaeva, Sergey Kovalchuk
YSC 2022


Unsupervised Learning to Analysis of Population of Models in Computational Electrophysiological Studies

Konstantin Ushenin, Nikita Detkov, Tatiana Nesterova, Dmitry Smarko, Artem Razumov, Olga Solovyova
IEEE SIBIRCON 2019

Projects

Master's thesis (2022)
Optimizing the training process of Vision Transformers in terms of required computations and data volume with the combination of Supervised and Self-Supervised learning

Person Verification with Super-Resolution (2019)
In this project we test the hypothesis of the possibility of improving the quality of Person Verification with a preliminary application of Super-Resolution

Bachelor's thesis (2020)
Research of influence of the label smoothing of pseudo-labeled data on training Convolutional Neural Networks

Own music recommendation system and singers' embeddings analysis (2019)
Collected the largest open source music lyrics dataset with wide metadata and trained NN for Multi-Class genre Classification followed by creation of singers' embeddings space to visualize their relations

Misc

My medium blog (2022)
Blog where I write short reviews on AI papers.
Telegram mirror is here.
Convolution with dilation from scratch (2020)
Implemented generalized 2D convolution with dilation from scratch in Python and NumPy.
Data-driven Astronomy Course (2020)
Since I'm in love with Astronomy and have some knowledge in Machine Learning, I recently started thinking about projects involving both of them. To start with, I completed this impressive course by The University of Sydney. Certificate.
Probabilistic Graphical Models Specialization (2018)
I was engaged in the translation of the second and the third courses from the specialization of probabilistic graphical models for further teaching the course on this topic at the university (author Daphne Koller has given her permission).
VK Group Bot (2018)

I created a bot for VK, which helps to cope with the flow of garbage in the newsfeed.
Kaggle Competitions Expert:

Other competitions:
  • 2nd place in the UrFU BEST Hackathon on Customer Churn Prediction (2019). Solution slides.
  • 3rd place in the “IT BEST COMPETITIONS” Hackathon on Late Payment Prediction (2018).