Every engineer uses AI now.
Hire the ones who use it well.

1. 01

   Can they brief an AI like they brief a junior?

   We give them a feature spec. They write the prompt they would actually send. We score for context, constraints, edge cases, and acceptance criteria, not for verbosity.

   A strong candidate response

   Implement a debounced search hook for the Postgres-backed /api/search endpoint we already use in SearchBar.tsx. 300ms debounce. Cancel in-flight requests on new input, using the AbortController we use elsewhere. Return { data, error, loading }. Do not introduce a new fetch library, we use native fetch. Handle the empty query by returning early without a request.

   Gives contextStates constraintsNames the edge case
2. 02

   Can they tell working from good?

   We show them AI-written code that runs. They explain what it does, flag the tells (an over-engineered class, a defensive try/catch swallowing real errors, a pattern nobody writes by hand), and say what they would change.

   class UserDataManager {
     private cache: Map<string, User | null>
     constructor() {
       this.cache = new Map()
     }
     async getUserById(id: string | null): Promise<User | null> {
       if (!id) return null
       try {
         if (this.cache.has(id)) return this.cache.get(id)!
         return await fetchUser(id)
       } catch (e) { return null }
     }
   }

   Candidate

   A class for what should be a function. It swallows errors silently, so the caller cannot tell a 500 from a missing user. And it never writes to the cache, so the cache never warms.

3. 03

   Can they fix one bug without touching anything else?

   We plant exactly one realistic bug in an AI-written function. They find it and patch it minimally. A broad refactor that misses the actual problem loses points.

   The candidate's diff

     function paginate(items, page, size) {
   - const start = page * size
   + const start = (page - 1) * size
       return items.slice(start, start + size)
     }

   Correct. One line, and no collateral refactor.
4. 04

   Can they catch every hallucination?

   We give them code with several planted flaws: invented APIs, off-by-ones, swallowed errors. We grade thoroughness. Did they catch all of them, or stop at the first one and call it good?

   Found by candidate, 3 of 3

   • lodash.deepFlatten does not exist. _.flattenDeep does.
   • catch (e) {} swallows the error. It should log or rethrow.
   • The loop runs in quadratic time. The outer pass should be a Set lookup.
5. 05

   Watch them work with the assistant.

   On the final question the candidate gets an AI sidebar inside the editor. Every prompt they send, every suggestion they accept, every chunk they reject, and every keystroke on top is recorded. The transcript goes to you.

   function debouncedSearch(query: string) {
     // accepted from the assistant
     if (!query) return
     if (controller) controller.abort()
     // candidate edit: was 200, made it 300
     timeout = setTimeout(...)
   }

   Sidebar transcript

   Candidate: use AbortController for cancellation

   Assistant: <draft>

   Candidate: debounce is wrong, it should be 300ms not 200ms

   4 prompts, 2 accepted, 1 rejected, 38% edited by hand

From a job description to a scored candidate, in one sitting.

1. 01

   Paste a job description.

   Or describe the role in one sentence. We pick up seniority, stack, and what the person will actually be doing.

2. 02

   Get an assessment in about 100 seconds.

   A custom test across all six dimensions, calibrated to the role.

3. 03

   Share a link. Get a scored report.

   Candidates take it async. You get per-question feedback, integrity flags, and for AI questions the full transcript.