Featured Projects
Skills and Qualifications
Languages
- Golang
- Python
- R
- Perl
- JavaScript
- TypeScript
- C/C++
Frameworks
- Node.js
Other
- ML/DL (computer vision
- NLP
- Data Science
- RL)
- architecture design
- Experimental Languages (Rust
- Elixir
- Scala
- Haskell
- Groovy
- D)
Projects
System for fixing safety violations
Data Scientist
The task was to create a system that detects whether or not a worker is wearing their helmet and gloves. If the system determines that the worker is not, it will automatically send a notification to the appropriate users about the violation. In order to be able to ascertain this, the system relies on video from surveillance cameras fed to the input in real time.
System of automatic allocation of rooms on the floor plan
Data Scientist
Anton needed to create a system that would facilitate the automatic search and extraction of room coordinates in drawings. The input is a picture of the drawing while the output is the coordinates of the rooms.
GPT-3-based service for text generation
Data Scientist
The task was to create a chatbot based on a regional version of the GPT-3 model from Sberbank. Anton undertook additional training before fine-tuning the model to meet the customer's requirements.
Recognition system for online poker table information
Data Scientist
Anton was tasked with creating a system to facilitate the automatic recognition of information relating to online poker games. The input was a video of 16 browser windows in which poker games were being played. It was necessary for the final system to divide the video into separate windows and analyze the core information within each one. Anton used OpenCV to segment the browser windows. He then improved the RCNN component that deals with segmentation of blocks on the poker table. Finally, he implemented a convolutional neural network to recognize all the cards on the table.
HARDI
Data Scientist
Anton participated in the development of an internal system for data analysis and visualization using R. The project team solved the existing data analysis problems and used Forecast Library to facilitate forecasting functionality. Graphs based on the data were built using Dygraphs Library.
International Heart Disease Diagnostic Challenge (Kaggle)
Data Scientist
The task was to predict the presence of one of four types of heart disease (or lack thereof) using data from the patient's physical examination. His work won the International Heart Disease Diagnostic Challenge at Kaggle.
Highload Cup (TOP 10)
Developer
A competition for developers specialized in high-load systems.
Algorithmic solutions in GoLang
Developer
Development of a SQLite database solution that allowed the load of up to 100 billion rows of data.
Searching for duplicates
Developer
The task was to build a system to help find duplicate images even though these images might have been distorted (cropped, rotated, changed color).