Hi there 👋

My name is Vadim Kirpa

Currently I'm working as a senior data scientist at Sberbank in Moscow.

I implement various machine learning algorithms for credit scoring process.

My tech stack consists of:

Languages: Python, C/C++
Tools: Pytorch, Scikit-Learn, Catboost, Lightgbm, Numpy, Pandas, PySpark, Greenplum, Matplotlib/Seaborn/Plotly

Welcome to my GitHub page!

My interests

Machine Learning (Classic ML, Deep Learning, NLP, CV, Reinforcement Learning)
Rust
Flutter for Mobile Development
Competitive Programming
Game Dev (Dipped my toes into Unity, now getting familiar with Bevy)
Web Dev (Had a somewhat experience with JavaScript and React)
Capture the Flag - I admire hackers because in order to hack something you have to understand how the whole system works, I think it's an enourmous feat
Human languages in general, especially Japanese - learning it on and off for about 10 years (more off than on though ;_;)

There are some projects I'm currenty wanting to make

The best Tower Defense game in the world
The first step has already been made, soon there should be a link to the repo
A website where I explain various Machine Learning concepts and algorithms with the best visualuzation and intuition possible (hope I'll get to that one someday)
A Phone App/Website for learning Japanese Kanji by suggesting the Next Best Kanji to learn through calculating some metric on a graph of Relations between Kanji and all Japanese words. Or put it another way - If a person already knows some set of Kanji, what Kanji should they learn to maximize the amount of new words that they can learn(meaning new words that don't require any new Kanji after the said one) after they learn said Kanji.