Hi, I am Ilya Kalimulin. My nickname is Ilirium. It is my personal blog. Currently, my family and I are living in 🇧🇷Brazil (UTC-3). My wife and I have 2 brilliant small kids. Before Brazil, we had the experience of living in 🇮🇳India (Goa), 🇪🇺European Union (🇱🇻Latvia), and 🇹🇷Turkey. I completed my Ph.D. degree in 2014 in the edge of microwave engineering and electronic design automation. Once I gave an oral speech at a scientific conference in San Jose (California, USA).
I have 12+ years in tech and analytics, of which 4+ years are working as a Data Scientist. From my experience, I am attracted to working with raster data stored in multidimensional arrays, like images and videos. I use digital filters, warping, interpolation, resampling, bitmasks, Boolean operations, LUT, removing backgrounds, etc., to process and generate images in geoimaging (satellite, weather, and environmental data). I gradually realized that all these operations refer to computer graphics and computer vision. Being involved in machine learning and deep learning projects, I was also always interested in trying to directly program the GPU and write my own implementation of a neural network from scratch.
To push my interest in CG generation, CV, and GPU programming beyond job tasks, I found the WGPU project and did some basic things with it with the Python wrapper and WGSL. Then I decided to work closer with WGPU and started to learn Rust (I am reading the book by Blandy/Orendorff/Tindall). Nicely, Rust and WGSL have some commons :) I am thinking about implementing 3 pet projects: a 2D raster engine (kind of “TinyDirec2D”), graphics and computer vision engine ( “TinySkia” + “TinyOpenCV”), and “TinyTorch” in Rust and WGSL on top of WebGPU.
I want to use this blog as a platform to showcase my experiments in the directions I have voiced and write articles & posts on technical topics of interest to me. At some point, I found that sharing my small steps motivates me to move forward. Equally important to me is sharing knowledge and experience with other people and the community through articles, as I have learned a lot from reading other people’s articles & posts.
Here are some more words about my background :) In the past, with several colleagues from the university, we applied our research to make a startup, which was a cloud-based simulation tool for electronics. As Data Scientist, I work with PyTorch, Numpy, and OpenCV. I developed a service to process weather radar images. I did experiments and developed MVP for pixel-classification of satellite images to predict the type of crops (CNN, RNN, LSTM).