LLM application building blocks

What every LLM system is assembled from: the model and its hosted APIs, prompt and context engineering, structured outputs and tool calling, retrieval over private data (two angles).

This module covers the building blocks that almost every LLM system is assembled from: the model and its hosted APIs (AWS Bedrock and the Claude API among them), the prompt and context around it, typed outputs and tools, and retrieval over private data. It moves from calling a model, to shaping what goes into and comes out of it, to grounding it in data. Retrieval is taught from two angles, the infrastructure for vector search and building with vector databases, and the two lessons sit side by side.

Lessons (6)

• Integrating hosted LLMs: AWS Bedrock, the Claude API and prompt engineering · Practice

  Hosted LLMs turn model serving into an API integration problem: the platform team does not run GPUs, but it owns access, quotas, cost, latency and the contract that application code uses.

• How LLMs work inside applications: tokens, context windows and model APIs · Practice

  A large language model is a function that takes a sequence of tokens and predicts the next one, repeated until a stop condition.

• Prompt engineering and context engineering · Practice

  Prompt engineering is the practice of writing instructions a model can follow reliably; context engineering is deciding what information occupies the window at each step and in what form.

• Structured outputs and tool calling · Practice

  Structured outputs make a model return data that conforms to a schema, so downstream code can consume it without parsing free text.

• Retrieval-augmented generation infrastructure and vector search · Practice

  Retrieval-augmented generation pairs a model's parametric knowledge with an external, searchable store, so answers can be grounded in current and private data and can cite it.

• Retrieval-augmented generation with vector databases · Practice

  Retrieval-augmented generation supplies a model with relevant passages from an external corpus at request time, so answers are grounded in data the model was never trained on and can cite it.

Changes

• Version 1, 2026-10-14: re-cut from the role pathway groups into a topic module (site 2026.10.2); no lesson changed