About

I am Konstantin Selivanov, a machine learning engineer and researcher focused on representation learning and deep learning. I have been the first machine learning engineer at several startups, helping build machine learning systems from early prototypes through production. Several of those companies - some backed by Google Ventures and Sequoia Capital - later achieved successful exits.

You can find me on LinkedIn and GitHub. Feel free to reach out.

Interests

My interests include representation learning (embeddings, contrastive learning, and information retrieval), natural language processing (large language models), generative models (autoencoders, normalizing flows, and diffusion models), and sequential models (RNNs, transformers, and state-space models).

I value minimalism and efficiency. My primary programming languages are Python, C, and Shell. I have used Linux for more than 20 years, and I remain deeply interested in mathematics and philosophy.