About

Hi, I’m Victor, a software engineer working on machine learning and performance engineering. This website is where I document my journey toward deeper expertise in machine learning systems and compilers.

You will find notes and articles that connect the mathematics of deep learning with its implementation: from backpropagation, model architectures, and PyTorch internals to distributed training, GPU kernels, Triton, MLIR, and compiler optimizations.

I am particularly interested in unpacking the intermediate steps that technical papers and documentation often omit—understanding not only how something works, but how it is represented, transformed, and ultimately executed efficiently on hardware.

These are working notes from an experienced software and ML engineer learning in public, revisiting fundamentals, experimenting with real systems, and gradually moving closer to the boundary between machine learning, compilers, and high-performance computing.