Dilshodbek Ikromov

Projects

• TinyTorch — A small deep learning framework built from scratch, one module at a time: tensors, autograd, layers, optimizers, attention, transformers, and then quantization and profiling to make it fast. [GitHub]
• Paper Implementations — Machine learning papers rebuilt in code to check I actually understand them. First one is the original Transformer, trained end to end on translation. [GitHub]
• Yandex Agents Week — Notebooks and notes from building LLM agents: tool calling written by hand, then the same agent rebuilt with LangGraph, with memory and a confirmation policy. [GitHub]
• Scientific Computing — Implementations and exercises from the Scientific Computing textbook. [GitHub] [Book]