AI Engineering Basics
APIs, retrieval and evaluation — building AI features the right way.
What you'll be able to do
- Explain how large language models work: tokens, limits and cost
- Call model APIs safely, with streaming, retries and structured output
- Build search and question answering over your own documents (RAG)
- Measure quality with evaluation sets instead of guessing
- Add guardrails, privacy protection and cost controls to an AI feature
About this course
Learn to build AI features the way engineers do: grounded in data, tested with evaluation sets, and kept safe and affordable. Projects run against recorded model responses, so grading never depends on a paid API.
Week 1 is free.
You need: Python basics (Python Foundations covers them) and a laptop or desktop.
Week by week
The planned syllabus. Lessons appear here as each week is published.
Week 1How LLMs work
Tokens, prompts, context limits and cost. Project: prompt experiments with recorded responses.
Week 2Calling a model API
Requests, streaming, errors and retries. Project: a command-line chat with logging.
Week 3Structured output and validation
Getting JSON back from a model and checking it. Project: a résumé parser that returns validated JSON.
Week 4Embeddings and semantic search
Turning text into vectors and searching by meaning. Project: search over lecture notes.
Week 5Retrieval-augmented generation
Chunking, retrieval, grounding and citations. Project: question answering over a syllabus.
Week 6Evaluation
Test sets, metrics and catching regressions. Project: an evaluation harness for Week 5.
Week 7Tools and function calling
Letting a model call your functions safely. Project: an assistant with calculator and search tools.
Week 8Safety, privacy and cost control
Guardrails, personal data and budgets. Project: guardrails and rate limits for your assistant.
Week 9A small AI web app
Serving an AI feature behind a simple web interface. Project: a deployed AI helper.
Week 10Capstone
Design, build and evaluate an AI feature. Project: an AI feature with an evaluation report.
Questions
- Is Week 1 really free?
- Yes. Sign in and do Week 1’s lessons and quiz without paying.
- Who scores my projects?
- A published rubric, applied the same way to every learner on your own GitHub repository. No AI decides your score, and paying never changes it.
- How long can I use it?
- It's self-paced: your schedule is built from the time you have each day, and you keep access for as long as we offer the course.
- What if it isn't for me?
- Ask within 7 days of paying for a full refund, no questions asked. Refund policy
Not sure where to start?
Answer ten quick questions and we'll build a plan that fits your time, your level and your goal.
Build my free plan