glossary
the terms of evidence-led hiring.
evidence card, skill map, mapping, staged situation and the other terms dwij uses, each defined in a sentence.
- evidence-led hiring
- a way of hiring engineers where every claim about a candidate is backed by something checkable, and each requirement of the role is judged on two questions: does the person understand it, and have they done it.
- evidence card
- what a founder receives for each candidate: every requirement of the role, marked on whether the candidate understands it and has done it, with the source behind every mark.
- skill map
- an engineer’s own record of the concepts they understand and the things they have done, each with the evidence behind it. the engineer sees it before any company does.
- mapping
- a 20 to 30 minute conversation, in english, by voice or typing, in which an engineer is asked about their real work through situations that unfold. it produces their skill map.
- staged situation
- a problem revealed one stage at a time, where each stage depends on the previous answer, so there is no single question to look up and the answers show how someone actually reasons.
- understands and has done it
- the two separate questions dwij asks about every requirement: can the person reason about the concept, and have they applied it in real work. a single score would blend them and hide the difference.
- not assessed
- a result on an evidence card meaning the conversation never reached that requirement. it is different from “not there”, which means it was asked about and wasn’t shown.
- role check
- a short follow-up to a mapping that asks only about the concepts a new role needs and the engineer’s skill map doesn’t yet cover.
- interview guide
- questions built from the gaps on a candidate’s evidence card, so a founder’s interview goes straight to what the evidence didn’t settle.
- ai fluency task
- an optional task, done in a sandbox with a coding agent dwij provides, that contains a trap the agent tends to fall into. it shows whether an engineer directs the agent well, checks its work and catches it when it is wrong.