← All courses

OpenAI Cookbook: A Practical Path for AI Application Developers

Developers who want to systematically learn the OpenAI API: from the first request to agentic systems. Analysts and ML engineers using LLMs for search, classification, and RAG. Team leads and architects who need to understand the capabilities, risks, and cost of LLM solutions. Level: intermediate. Solid Python required; ML experience is not necessary.

Log in to start →

Course preview

What you will learn

Explore the full curriculum before you enrol. Lessons unlock after purchase or free enrolment.

01

API Fundamentals and Prompting

Topics covered

  • Understand how the Chat/Responses API works: message roles, parameters, response format.
  • Learn to count tokens and estimate request costs before sending them.
  • Master streaming responses and Structured Outputs.
1 lesson10 questions~25 min
02

Embeddings and Semantic Search

Topics covered

  • Understand what vector text representations are and why they "understand meaning."
  • Learn to get embeddings via the API and compare texts using cosine similarity.
  • Apply embeddings to four classes of tasks: search, classification, clustering, recommendations.
1 lesson10 questions~25 min
03

RAG — Retrieval-Augmented Generation

Topics covered

  • Understand the RAG architecture and when it's needed instead of fine-tuning or a long context.
  • Build the full pipeline: document preparation → chunking → indexing → retrieval → generation.
  • Master parsing "messy" sources (PDFs) and managed File Search in the Responses API.
1 lesson10 questions~25 min
04

Function Calling and Agents

Topics covered

  • Understand the function calling mechanism: how the model "calls" your code and what actually happens.
  • Learn to describe tools with JSON schemas and build a tool-use loop.
  • Master agent patterns: the "think → tool → result" loop, handoffs between agents, parallelism.
1 lesson10 questions~25 min
05

Images and Audio

Topics covered

  • Master image generation and editing (GPT Image, DALL·E).
  • Learn to "read" images with vision models: captioning, tagging, data extraction.
  • Understand the technique of analyzing video via frame sampling.
1 lesson10 questions~25 min
06

Fine-Tuning — Adapting Models

Topics covered

  • Understand when fine-tuning is justified, and when prompting or RAG is enough.
  • Master preparing and validating training data in chat format.
  • Go through the full cycle: uploading data → running a job → monitoring → using the model.
1 lesson10 questions~25 min
07

Evaluation and Reliability

Topics covered

  • Understand why systematic evaluations (evals) are needed and how to build them.
  • Master LLM-as-a-Judge: when a model can evaluate a model, and how to trust that judge.
  • Learn to build guardrails: input and output checks, protection against hallucinations.
1 lesson10 questions~25 min
08

Cost Optimization and Production

Topics covered

  • Learn to work reliably under rate limits: retries, backoff, parallelization.
  • Master prompt caching and structuring prompts for the cache.
  • Use the Batch API for offline tasks at a 50% discount.
1 lesson10 questions~25 min
09

Final Exam

Final exam
35 questions~60 min