Konstantin Selivanov

I am a machine learning engineer and researcher focused on representation learning, information retrieval, generative models, and sequential models. I like compact ideas, efficient systems, and explanations that connect mathematical intuition to working code.

Writing

• Fano’s inequality - why features must be informative - A short information-theoretic view of what a useful representation must preserve.

• Two ways to implement the Game of Life - A comparison of object-heavy and array-oriented representations in Python.

Open source

• Transductive regular expressions - A C99 implementation of transductive regular expressions.

• Brzozowski derivatives - A Python implementation of regular-expression derivatives.

• Finite-state transducer library - A finite-state transducer library written in C99.