Curriculum
7 modules, from tokens and sampling to production AI systems. 7 are open with 40 lessons; 0 are in development.
Module 01
Model Foundations
Tokenization, embeddings, attention, sampling, context windows, and the KV cache.
6 lessons · 46 topics · about 2 h
Module 02
Inference and Model Optimization
Prefill versus decode, latency and throughput, quantization, batching, and cost per request.
6 lessons · 42 topics · about 2 h
Module 03
Prompt and Context Engineering
Instruction hierarchy, structured outputs, context selection, prompt regression tests, and injection.
6 lessons · 43 topics · about 2 h
Module 04
Evaluation and Experimentation
Success criteria, metrics, LLM-as-judge limits, baselines, ablations, and error analysis.
5 lessons · 40 topics · about 2 h
Module 05
Retrieval and Knowledge Systems
Chunking, dense, sparse and hybrid retrieval, reranking, grounding, and RAG failure analysis.
5 lessons · 35 topics · about 2 h
Module 06
Agents and Harness Engineering
Tool calling, control flow, state, retries, traces, termination conditions, and bounded autonomy.
6 lessons · 42 topics · about 2 h
Module 07
Production AI Systems
Routing, fallbacks, budgets, caching, observability, safety boundaries, and when to fine-tune.
6 lessons · 40 topics · about 2 h