FRONTIER AI MAP
The topics that define current AI systems, each with the ideas to master, the skills you will gain, where to learn it here and the primary sources.
"Planned" means no published course covers the topic yet. The reading list is available today.
Tokens, embeddings, attention and scaling: the mechanics underneath every current model.
Prompting, structured output and retrieval-augmented generation for applications grounded in your own data.
Tool-using models, agent loops, the Model Context Protocol and how to keep autonomy bounded.
Chain-of-thought, test-time compute and process supervision: why newer models think before answering.
Instruction tuning, RLHF, DPO and parameter-efficient methods such as LoRA.
Benchmarks, task-specific evals, LLM-as-judge and red-teaming, plus why leaderboards mislead.
Vision-language models, image and audio understanding, and building apps that mix modalities.
Alignment methods, misuse risks, privacy, prompt injection and the human oversight that real deployments need.