Free Lesson
Prompting Like an Engineer, Not a Wizard
30 min
Jul 7, 2026 2:00 PM
Virtual (Zoom)
In this video
What you'll learn
Write prompts as contracts, not incantations
Turn vague instructions into an explicit contract, fixed inputs, output shape, and constraints, that you can test.
Get structured output you can trust
Make the model return a defined schema with typed fields, then validate it, so code never parses free-form text.
Constrain the input before it hits the model
Never pass raw user text straight through; shape, bound, and sanitise it first.
Why this topic matters
Most 'flaky' LLM features aren't the model being random, they're a vague prompt it was free to read three different ways. Prompting like an engineer means writing an instruction it can't misread, getting output your code can trust, and constraining the input before it ever reaches the model. In 30 minutes you'll turn prompts from hopeful prose into something you can actually test.
