Lessons
Interactive lessons, organised by track. Each track builds toward a working mental model of one slice of production AI.
Agentic Systems
4 lessonsWorkflow first; agent only where decision flexibility is the value. Decision matrix, bounded-agent trace playground, failure-mode heatmap, safety audit. Layer 10.
LLM Foundations
4 lessonsWords to numbers and back. The mechanics every other layer assumes. Layer 1 of the systems-engineering track.
Production Architecture
4 lessonsThe twelve components, observability, cost, prompt injection. What separates an LLM demo from a system the business can rely on. Layer 12.
RAG Systems
5 lessonsChunking, retrieval, reranking, parent-child retrieval, diagnosis. The discipline of giving the model the right context. Layer 9.