Building AI Apps with LLMs

What you’ll learn

• Understand the model

  Reason about tokens, context windows, and sampling instead of treating the model as a black box.

• Get structured output

  Make a model return data your program can rely on, and validate it before you use it.

• Wire up tool calling

  Let the model call your functions, and keep the loop bounded and observable.

• Ship it affordably

  Stream responses, cache what repeats, and keep latency and spend under control.

1. 1LLM FundamentalsWhat the model is doing, what it costs, and which knobs actually change the output.26m 39s

   1. Lesson 1.11.1Tokens and context windowsFree preview
      11m 31s
   2. Lesson 1.21.2Temperature and sampling
      8m 11s
   3. Lesson 1.31.3Choosing a model for the job
      6m 57s
2. 2Prompting That Holds UpWriting prompts that behave the same way on the thousandth call as they did on the first.34m 47s

   1. Lesson 2.12.1Designing the system prompt
      21m 1s
   2. Lesson 2.22.2Few-shot examples
      4m 31s
   3. Lesson 2.32.3Structured output and JSON
      9m 15s
3. 3Tool Calling and AgentsLetting the model act: exposing functions, running the loop, and keeping the whole thing bounded.54m 22s

   1. Lesson 3.13.1Tool calling fundamentals
      31m 14s
   2. Lesson 3.23.2Building an agent loop
      15m 50s
   3. Lesson 3.33.3Guardrails and evaluation
      7m 18s
4. 4Shipping to ProductionThe engineering around the model: streaming, cost, caching, and behaving well under load.31m 27s

   1. Lesson 4.14.1Streaming responses to the browser
      8m 56s
   2. Lesson 4.24.2Managing cost and latency
      13m 25s
   3. Lesson 4.34.3Caching and rate limits
      9m 6s