Skip to main content

Artificial Intelligence

These notes are for individual developers who want AI to write code and do work alongside them. Groups 1–7 follow the order in which you meet the work; group 8 is a reference shelf for readers who train their own models.

Choose a Group by the Job​

StepTopicYour job
1Coding with AI agentsPick a coding agent, work out what it costs, and make it reliable in your repository.
2Models, APIs and plansChoose a model tier by task, then a vendor; read prices and vendor scores critically.
3Building agentsAdd a tool, a Skill or an MCP server to the agent you use, or write your own loop.
4Retrieval, context and memoryDecide which material reaches the context and what persists across sessions.
5Evaluation and evidenceTest a model or system on your own task, and judge what other people’s scores show.
6Running locallyDecide whether to run a model on your machine at all, then pick a setup that fits and is fast enough.
7How LLMs workFollow one prompt through tokens, decoding, attention, the KV cache and post-training.
8ML and deep-learning foundationsReference shelf: classical learning, neural building blocks, generative models and data, for readers who train their own models.

For prerequisites, see the mathematics map and the computing map; look up terms in the data science vocabulary.

Explore connectionsOpen network