About Me
I'm an engineer proficient in Python, C/C++, and frameworks like Tensorflow and Keras. With experience in Machine Learning, I've utilized CNNs, Tensorflow, and Pandas to enhance data processing efficiency. Projects like TailTrail and BookBinder demonstrate my expertise in AngularJS, Supabase, and Symfony. Fluent in English, Russian, and Kazakh, I excel not only in development but also in fostering collaborative environments. Committed to continuous learning and passionate about tackling challenges, I'm ready to contribute to the team.
- Keras
- Numpy
- Tensorflow
- Pandas
- Matplotlib
- OpenCV
- Python
- C/C++
- Java
- JavaScript
- Next.js
- Angular.js
- PHP
- SQL
- Git
Professional Experience
Formula Electric
TensorflowCNNPandasNumpyMatplotlib
- Researched deep-learning architectures used to process point-cloud data. In conclusion, decided on PointNet++
- Performed data collection for the creation of custom-tailored dataset for cone detection on race tracks
- Provided insight into point-cloud data which increased performance by 19.3%
MinDCeT NV
Python
- Extended in-house library for microchip testing with electronic equipment
- Developed Python code for autonomous test of microchip
- Streamlined the process and achieved an increase in efficiency of 10%
R&D
Master's Thesis: Compare RT-ST-GCN with RT-Shift-GCN++ on hardware
PyTorchCNNGCN
- Researched papers ST-GCN and Shift-GCN and related
- Experimented with RT-ST-GCN model configuration to point out the parameters which has the impact on performance and hardware efficiency
- Next Steps:
- Transform ShiftGCN++ to RT-ShiftGCN++
- Train both models with different parameters and compare the accuracies
- Compare the efficiency and performance on end-devices
My Projects
TailTrail
Angular.JSSupabaseVercelGoogle Analytics
- Developed with front-end AngularJS, back-end Supabase, and deployed with Vercel
- Integrated with basic Google Analytics for accumulation of user metrics for future improvements
- Sped up user flow from 5.6 mins to 2.4 mins by introducing recommender system into the search measusecondary using Hick's Law
Handwriting Recognition
Deep LearningTensorflowKerasPythonOpenCVNumpyCNN
- Performed image multiclass classification task by leveraging OpenCV image processing techniques and Tensorflow framework
- Increased model's accuracy by 42.3% by introducing CNN-based model
- Provided insight on the most misclassified letters using color-coded heatmap
Qt Game
C++QtCreatorUML
- Made a game in a team using QtCreator IDE and C++
- Designed extensible architecture for a renderering with a Sprite, Text, Color rendering options already available
- Added multiple animation choices running in parallel with QParallelAnimationGroup which are also extensible
BookBinder
HTMLPHPCSSSymfonyMySQLJavaScript
- Based on PHP back-end, and CSS with Symfony framework for front-end
Simple Notes
FlaskPythonHTMLCSSSQLite3
- Simple application to manage notes built with Flask and SQLite3 relational database