Learn · Practise · Follow the field
Find your place in AI evaluation.
Curious about evals? Start with what the work looks like. Bring your experience, learn how to judge AI systems, and discover where your skills could fit.
New to evals? Start here.
An evaluation checks whether an AI system meets defined expectations. Someone has to decide what good looks like, create realistic tests and investigate failures. That work brings together engineering, subject knowledge and careful judgment.
Choose your next step
I don't code
Explore human evaluation, domain expertise and operations. Learn which skills transfer and what employers actually ask for.
Explore work without a coding requirement →I build software
Connect tasks, environments, graders and traces. Start with a small reproducible test before building infrastructure.
Understand the components →I want to try the work
Create a rubric, judge fictional support responses and explain your decisions. No paid tool or sign-up required.
Build a first portfolio exercise →I want to keep up
Read research, useful tools and developments across the field, with original sources and clear explanations.
Read the Evals Brief →New: evaluate an AI sales report
The totals are correct. The explanation is not. Use a four-deal example to practise checking numbers, evidence and management actions.
Bring the skills you already have
Writing clear instructions, spotting factual errors, understanding a workflow and explaining a difficult judgment are useful starting points. They become stronger evidence when you can apply them consistently to a documented test.
Explore career paths · Meet the companies · Browse current jobs