A Hands-On Build Guide: Pure-Python From-Scratch LLM Decoding Engine
A step‑by‑step guide to building a pure‑Python LLM decoding engine from scratch, complete with runnable code, temperature, top‑k, top‑p, and typical sampling, perfect for a CV.
Llm
Dynamic Prompt Compression System with Learned Importance Scores
A hands‑on guide to building a from‑scratch prompt compression system that learns token importance and truncates context for fixed‑size LLM windows.
Continuous Batching LLM Scheduler: Hands‑On Build Guide
A step‑by‑step guide to building a runnable continuous batching LLM scheduler that you can showcase on your CV, with architecture, code, testing, and extension ideas.
Building a Pure-Python Top-p/Top-k Sampler with Temperature Scaling for LLM Logits
This post walks through building a pure-Python top-p/top-k sampler with temperature scaling, providing runnable code and a roadmap for productionizing it.
Lazy Loading LLM Weights with Safetensors, Async Prefetch, and Checksum Verification
A practical guide to building a lazy weight loader for large LLMs using safetensors, async prefetch, and checksum verification.