Prompt Engineering Course
Write prompts that get consistent, useful results from language models, with clear patterns and practical examples.
26 lessons · about 6 hours of reading · Free
Start lesson 1Course syllabus26 Lessons
What Is Prompt Engineering? A Practical Beginner's Guide
Learn what prompts do, how prompt engineering differs from model training and retrieval, and how to improve results by testing against clear goals.
How a Model Reads Your Prompt: Tokens and Context
See how text becomes tokens, how a model generates a reply, and why context windows, model weights, and generation settings affect what you get.
Choosing a Model and Reading Its Documentation
Choose a language model by testing it on your task, then use its official model page, documentation, and changelog to check what it supports.
Controlling Model Output: Sampling and Limits
Learn what temperature, top-p, top-k, token limits, stop sequences, and repetition penalties do, and why support varies by model.
Your First Prompt Workflow: Test One Change at a Time
Build a small prompt experiment: define success, save a baseline, change one thing, and compare several runs fairly.
Be Clear and Direct: Give the Model a Usable Task
Turn vague requests into clear instructions by naming the task, audience, needed details, and shape of the answer.
System, Role, and Context Prompts: Give Instructions Their Place
Separate stable behavior from the current request, use role framing carefully, and add only context that helps the task.
Zero-Shot, One-Shot, and Few-Shot Prompting
Learn when examples help, how to choose representative cases, and how to avoid teaching the model the wrong pattern.
Structure Prompts with Delimiters, Tags, and Templates
Separate instructions from input, use clear labels, and build reusable templates that fail clearly when data is missing.
Structured Output: Get Answers You Can Parse and Check
Choose an output format, define the fields you need, validate the response, and handle failures deliberately.
Prompting with Long Documents and Large Context
Organize long documents, ask evidence-based questions, summarize in stages, and choose when retrieval beats pasting.
Prompt Recipes for Ten Common Jobs
Start with practical prompts for summarizing, classifying, extracting, rewriting, translating, and more, then test each one.
Chain-of-Thought Prompts and Thinking Models
Learn when step-by-step prompting can help, what reasoning models do differently, and why a plausible explanation is not proof.
Step-Back Prompting and Prompt Chaining
Use a broader question to find a helpful framework, then split longer tasks into steps with checks between them.
Self-Consistency, Tree of Thoughts, and ReAct
Compare three advanced ways to handle difficult tasks, understand their costs, and choose the simplest approach that works.
Self-Evaluation and Reliability Checks
Ask a model to review its answer, understand what confidence can tell you, and choose checks that can catch real errors.
Define Success and Build Test Cases
Turn “good enough” into clear checks, then build a small test set that covers routine inputs, edge cases, and valid refusals.
Compare Prompts with Graders
Run prompt versions on the same cases, use checks that fit the task, and read results without mistaking a small sample for proof.
Debug a Failing Prompt
Diagnose whether a failure comes from unclear instructions, missing context, conflicting rules, format drift, or a task beyond the model.
Automatic Prompt Engineering
Use a model to propose prompt improvements, compare candidates against your cases, and keep human judgment in the loop.
Version and Document Your Prompts
Keep prompts reviewable, record why they changed, run regression checks, and make production updates traceable and reversible.
Hallucinations and Grounding
Reduce unsupported answers by providing evidence, allowing uncertainty, asking for traceable quotes, and checking claims against their sources.
Bias and Debiasing Prompts
Find unfair patterns in model responses with controlled prompt comparisons, balanced examples, and checks that reflect the task.
Prompt Injection and Safe Prompting
Keep untrusted text in its place, check model outputs, and limit what a compromised prompt can cause.
Prompts in Production
Keep prompts reliable as real users depend on them by measuring quality, cost, and latency, testing changes, and planning recovery.
Capstone: Build and Test a Prompt Library
Bring the course together by defining one task, comparing prompts, checking a winner, and recording it for future changes.
Related courses
Get Prompt Engineering by email
One lesson a week, in order. Free, and you can unsubscribe from any email.
Frequently asked questions
Who is the Prompt Engineering course for?
Anyone who uses a language model for real work: beginners who only used a chat window, writers and analysts, and developers who put prompts inside applications.
What do I need before starting the Prompt Engineering course?
None. You only need access to a language model to try the examples.
What will I learn in the Prompt Engineering course?
You will build and test a prompt library, so you can say a prompt passes a set of checks instead of just seems better.
How long does the Prompt Engineering course take?
It has 26 lessons and about 6 hours of reading, before you spend time on the practice tasks. Go at your own pace; the lessons are written to be done in order.
Is the Prompt Engineering course free?
Yes. Every lesson is free to read, and you do not need to sign in.