Prorated

Hiring / Founding & AI engineers

Hire engineers who can actually build.

We source founding and AI engineers, then assess them on realistic work, with AI allowed. You get a shortlist backed by results, recorded decisions, and comparisons against independent AI runs.

Start a hireSee a sample challenge

Example benchmark

Requirement met, by taskshare of runs, %
  • Model A
  • Model B
  • Model C
  • Model D
  • Candidates with AI

OPS-201 · Conflicting claim owners

  • 70%Model A
  • 60%Model B
  • 50%Model C
  • 40%Model D
  • 62%Candidates with AI

OPS-202 · Reply sent twice

  • 30%Model A
  • 20%Model B
  • 20%Model C
  • 10%Model D
  • 48%Candidates with AI

OPS-203 · Missing timeline event

  • 80%Model A
  • 70%Model B
  • 60%Model C
  • 60%Model D
  • 71%Candidates with AI
  1. Reproduced. A retry after a lost response creates reply
  2. Verified. Replayed send produces one reply and one email.

From role to shortlist.

  1. Define the workTell us what this engineer will own. We turn it into a role-specific evaluation.
  2. Find candidatesWe source engineers and can assess applicants already in your pipeline.
  3. Watch them workCandidates investigate real engineering problems. AI tools are allowed.
  4. Compare the evidenceReview their results, process, and performance alongside AI-only runs and other candidates.

What we score, and what you can inspect.

A passing test isn't the whole story. Each rating links back to what the candidate actually did.

The five measures in a Prorated report
MeasureWhat it checks
Delivered resultDoes the solution meet the requirements?
Diagnosis & validationDid they find the cause and verify the fix?
Priorities & tradeoffsDid they make sensible choices under constraints?
AI supervisionDid they direct, question, and correct AI?
Handoff & honestyDoes the final report match the work?

Sample report

OPS-202 · Candidate 114rubric score, 0 to 4
  • Candidate
  • Peer median
  • AI-only median
  • Delivered result3.2
  • Diagnosis & validation3.6
  • Priorities & tradeoffs3.0
  • AI supervision3.4
  • Handoff & honesty3.5
  • Reproduction loggedSecond POST after a timeout created reply
  • Fix verifiedReplayed the send. One reply, one email, test reply_once passed.

How the benchmark works

For early teams making critical hires.

  1. Founding engineersPeople who can own ambiguous product and infrastructure work end to end.
  2. AI engineersPeople building AI-native products, agents, and production integrations.

Common questions.

Do you source candidates or assess ours?

Both. We can find engineers for your role, evaluate your applicants, or combine the two.

Can candidates use AI?

Yes. We look at what they build and how they direct, check, and correct AI.

Does the AI benchmark decide who gets hired?

No. It provides a reference for the same challenge. Your team reviews the evidence and makes the hiring decision.

How does pricing work?

It depends on the role and whether you need sourcing, assessment, or both. Contact us to scope the engagement.

Hiring a founding or AI engineer?

Tell us what you need. We'll show you how we'd source and assess the role.

Start a hire