rebaseBuild in context.
Make a scoped change inside an existing codebase. Understand the architecture and product constraints before touching the code.
COMPREHENSION + DELIVERY A new signal for engineering hiring IN DEVELOPMENT
AI can write code.
Can they own it?
Meet the engineer behind the output. Assess how candidates build, question and verify with AI in real codebases.
For teams hiring engineers who think beyond the green check.
01
export async function handleWebhook(event) {02
const key = event.id;03
+ const claimed = await db.claim(key);04
+ if (!claimed) return { status: 200 };05
await createOrder(event);06
return { status: 200 };07
}08
09
// What happens if createOrder fails?VERIFICATION2 passed · 1 to investigate
✓ Duplicate delivery produces one order
✓ Existing behavior is preserved
? Recovery after a partial failure
⌁ Existing codebases✳ AI welcome◈ Evidence you can inspect
01 / THE WORKCloser to the job.
Closer to the job.
Closer to the truth.
Good engineering takes more than a working implementation. Give candidates the kind of work they’ll actually do.
Review what matters.
Review a realistic pull request. Catch consequential issues, explain the customer impact and decide whether it should merge.
CORRECTNESS + JUDGMENTFollow the failure.
Reproduce a bug, investigate the cause and verify the repair. Show the evidence that supports the diagnosis.
DIAGNOSIS + VERIFICATIONPressure-test a design.
Use AI to explore architecture options. Defend the tradeoffs and challenge the edge cases before calling it done.
TRADEOFFS + OWNERSHIPTwo planned formats: a practical codebase assessment and a separate AI-assisted system-design exercise. Choose the fit for the role.
02 / THE SIGNALThe diff is only
The diff is only
part of the story.
See what a candidate understood, which AI suggestions they challenged, what they tested and where uncertainty remains.
Help shape the first assessmentsEXAMPLE EVIDENCE NOTEVerification
“The candidate identified a permissions regression and added a cross-account test. Their repair handles sequential retries, but overlapping retries remain unverified.”
Code changesTest resultsAI interactions
A useful follow-up. A concrete basis for judgment.
03 / THE FIRST COHORTBuild the next hiring
Build the next hiring
signal with us.
We’re inviting actively hiring startup teams to become early partners, try the first assessments and help shape the evidence.
Starting with recent YC batches. One selected exercise per candidate, at $10 per assessment.
STARTEREARLY PARTNER
$100
10 candidates
A small first cohort.
Find your hiring signal.
GROWTHEARLY PARTNER
$500
50 candidates
A wider candidate pipeline.
Keep the evaluation consistent.
ENTERPRISEEARLY PARTNER
$1,000
100 candidates
A larger hiring round.
Evidence for every candidate.
Planned pilot pricing, with an AI allowance agreed before starting. Register your interest—no payment today.
EARLY ACCESS / REBASEHire for the work
Hire for the work
that comes next.
Join the waitlist for launch updates and a chance to shape the first pilots.
YOU’RE ON THE LISTGood work starts
Good work starts
with good judgment.
We’ve saved your interest. We’ll be in touch when early access is ready.
Can candidates use AI?
Yes. AI is part of the assessment. We’re interested in how candidates direct it, evaluate its suggestions, verify the result and explain their decisions. Elaborate prompts alone don’t demonstrate competence.
What will hiring teams receive?
A concise evidence report linked to the actual work: code changes, review findings, test results, design decisions and relevant AI interactions. The goal is specific observations and useful follow-ups for your hiring team.
Is Rebase available now?
Rebase is in development. The waitlist is open for hiring teams and engineers. We’re preparing the first paid startup pilots; joining the list doesn’t require payment or guarantee pilot access.
Does Rebase make the hiring decision?
The hiring team makes the decision. Rebase is being built to show what was observed and what remains unverified, giving you concrete questions to explore in a follow-up conversation.