// TODO: I’m martinuke0
Welcome to my corner of the internet. This website is a personal blog which I use as a platform to document my learning journey and showcase it for the world to see.
Build a continuous batching LLM inference engine with paged attention and block-based KV cache in pure Python
A hands-on guide to implementing a continuous batching LLM inference engine with paged attention and block-based KV cache in pure Python.
BPE Vocabulary Compression via Optimal Merge Scheduling
Learn to build a BPE tokenizer that compresses vocabulary by scheduling merges optimally, with runnable Python code and a roadmap to production.
Hands‑On Build Guide: Group‑wise INT4 Weight‑only Quantization with Per‑channel Scales and Per‑token INT8 Dynamic Activation Scaling for LLM Inference
A step‑by‑step guide to implementing group‑wise INT4 weight quantization and per‑token INT8 activation scaling from scratch in pure Python, ready for LLM inference.
Pure‑Python LoRA Adapter: A Portfolio Project for Systems Engineers
A hands‑on guide to implementing a pure‑Python LoRA adapter and fine‑tuning pipeline, ideal for building a portfolio project that signals real systems expertise.
From‑Scratch GGUF and safetensors Binary Weight Loader with Memory‑Mapped Q4_0 Dequantization in Pure Python
A minimal pure‑Python loader that memory‑maps Q4_0 quantized weights from GGUF and safetensors files, ready for inference or further processing.