LLM API Mastery

Ten minutes of environment work so every code snippet in this course actually runs: provider accounts, API keys, a Python virtualenv, both SDKs, and a smoke-test call to prove the plumbing before Lesson 1 assumes it.

The single most important fact in agent engineering: the model is stateless. A 'conversation' is you resending an ever-growing array. Every agent pattern you'll ever build follows from this.

How a token actually gets picked, step by step — logits, softmax, and the sampling parameters that shape the draw — and how the dials changed. Classic sampling (temperature, top_p) still runs most of the industry, but 2026 frontier models replaced those knobs with adaptive thinking and an effort parameter. Streaming turns dead air into perceived speed.

The mechanism that turns a text generator into something that can act. Crucial mental model: the model never executes anything — it emits structured JSON, and your code does the work.

When you need data, not prose: forcing model output to conform to a schema, and what to do when it doesn't.

An agent lives or dies on the unhappy path. Rate limits, timeouts, overloaded servers, refusals, context overflows — production behavior is defined by how you handle these.

Real agent tasks aren't text-only: screenshots in bug reports, invoices as PDFs, documents to extract from. Multimodal input is just more content-block types in the same messages array.