AI engineering fundamentals

What you will be able to do

Skills you add to your Herizon profile when you finish: AI Engineering, LLM APIs, AI Evaluation.

Who it is for

Developers and career switchers who can write a small program and want to build with language models: junior developers, data analysts who code, and technical founders adding AI to a product. It is not about training models, which is what the machine learning fundamentals course covers.

The situation you work through

Ronja, a junior developer at Kaluvo, a shift-scheduling app for small restaurants in Tampere with 9 people, builds a draft-reply assistant for 1,400 support emails a month: a first model call, a prompt checked by code, retrieval over 140 help articles, an eval set of 60 tickets that scores 27, 46, 49 and then 52 against a bar of 50, guardrails, one read-only tool and a flagged rollout to a support team of 3.

Your practice task

Build one small AI feature that answers questions from documents you are allowed to use. Write 20 eval cases with their checks and a ship bar before you write the prompt, including 3 adversarial cases. Call a model API from server code, add retrieval over the documents, and run the eval on two versions. Report the passes for each, tokens per call, a monthly cost estimate from today's prices, and the median and p95 latency. Add one input rule and one output rule with a case that tests each, and write a rollout plan with a flag and an off switch. Submit one page and a repository link without keys or personal data.

What is inside

1. What AI engineering is
2. Your first model call
3. The prompt is a spec
4. Tokens, cost and latency
5. Retrieval over your own docs
6. An eval set before opinions

and 6 more pages

Finish with a short quiz to earn your certificate.