A learning library written as much for myself as for anyone else. Each topic is a single coherent document, readable end-to-end.
Foundations
SWE + LLM essentials for the AI era. Starting from a 'how AI got here' time anchor, build the complete baseline — vocabulary, engineering substrate, system judgment — before specializing in system design / data engineering / model / Agent / etc.
Agent Engineering
Harness engineering — turning a model into a reliable production agent. From a 'how agent engineering got here' time anchor, build a runnable, safe, hijack-resistant harness, make it survive long tasks, then learn to measure, read real systems, and extrapolate. Prereq: Foundations (#1).
System Design
Serving trained models at production scale. From a 'how AI systems got here' time anchor: the distributed substrate (parallelism + collective communication), the inference-serving spine (KV cache + batching + clusters), performance & hardware (roofline + quantization), reliability ops, and operating agents as a production service at scale. Prereq: Foundations (#1).