Our Hiring Process.

Lock the four success pictures first.

Before sourcing opens, we lock the picture of what a successful hire looks like at each milestone. The JD is built backwards from these. If we can't write them, the role isn't well-defined yet — and we shouldn't be hiring.

• 6 months — first wins, ramp complete, owns clear scope.
• 12 months — ships independently in their domain.
• 18 months — leads projects, raises the bar for the function.
• 24 months — multiplier effect (hires, mentors, lifts peers).

Kolbe A index expectation.

Every role has a target Kolbe A profile written into the JD. Candidates take the Kolbe assessment as part of the process (PerkUp pays). If a candidate's Kolbe is way off the target, we surface that explicitly to them before either party invests more time.

Three things must be true before sourcing kicks off. Every role, no exceptions.

• Hiring channel created. One Slack channel per active role in the format #hiring-<role>. Used for candidate notes, interview debriefs, and offer decisions.
• Posting destinations chosen. We are comfortable paying for ads on platforms that produce candidates. We are not comfortable paying for platforms with patterns of fraud or fake candidates — every paid platform gets a quick due-diligence check before money goes out.
• Kickoff date set. Sourcing starts on a specific date so funnel rates are measurable from a known baseline.

Cast as wide as paid budget allows. Every candidate gets added to the Roles database with clean status discipline — Sourced → Screening → Interviewing → Trial → Hired / Rejected.

Wide-net sourcing only works if we're responsive. The communication SLA (TBD) is how we keep the funnel honest — candidates we ignore stop applying, and the platforms we pay for stop sending great people.

Every Test Gorilla assessment we send must contain exactly two things:

• 5 skill assessments drawn from the Test Gorilla library that match the role.
• 5 custom questions written specifically for the role — and these must be Claude-generated from the JD, not hand-written from scratch. Hiring Manager reviews and approves before sending.

Why Claude-generated custom questions.

The JD already encodes what good looks like. Deriving the custom questions from the JD (instead of inventing them in parallel) does three things at once: it removes the bottleneck of "we're waiting on the HM to write good questions", it keeps the funnel moving, and it keeps the question set tightly aligned with the success criteria from Stage 0. Hiring Manager review still catches anything off-base.

Candidates whose Test Gorilla score lands in our acceptable band are asked to submit a 5-minute Loom covering four prompts:

1. Their experience — a quick walk-through of the most relevant past work.
2. Why they're interested in the role — what specifically about this job attracts them.
3. Why they're interested in PerkUp — what they know about us, why this company over others.
4. Why they think they're a good fit — their own pitch for the role.

No slides required. We're explicitly testing how someone communicates async with minimal prep — that's a real-world job skill at PerkUp, not a hoop. The Hiring Manager reviews the Loom and decides if the candidate advances.

The first synchronous conversation. Just the Hiring Manager and the candidate — one PerkUp person in the room, no exceptions. Confirms what came through in the Loom and surfaces anything that needs deeper exploration. If it goes well, we advance to Stage 6. If not, we end here cleanly and respectfully.

A deeper interview led by the Hiring Manager. After this stage, the HM starts handing the candidate off to the rest of the hiring committee in additional one-on-ones. Each committee session is still exactly one PerkUp person at a time.

The candidate shares their screen and the Hiring Manager runs role-specific tests:

• Engineering — pair on a real problem from the actual backlog.
• Design — critique a real artifact or sketch a real flow live.
• Operations / CS — work through a real customer scenario end-to-end.
• Marketing / Brand — talk through a real campaign decision and edit a live asset.

The goal is real-time work signal — what does this person do when something doesn't go to plan, how do they think through a constraint, how do they communicate decisions in motion. The screen share is the source of truth; what they say is secondary to what they do.

The take-home is always paid. The pay tier depends on the role's target salary band.

Scope is a real task we'd actually need done — never invented busywork. Sized to fit inside 1–3 hours so candidates with day jobs aren't penalized. We pay the take-home rate even if it doesn't lead to an offer — that policy is what makes it ethical to ask for the work in the first place.

At least one reference check is required for every hire. For senior or strategic roles — anything where a bad call would meaningfully change the trajectory of a function — we do two. Run by the Hiring Manager after the take-home result is in and before the offer call.

Default reference questions live in the cross-role question bank; per-role pages can override or add. The HM logs the reference call in the hiring channel so the committee has visibility before the offer goes out.

The offer is always made on a live call, never by message. On the call we confirm four things explicitly, in this order:

1. The salary amount, verbally. No ambiguity, no email-only quote.
2. Engagement is via Deel as a contractor. The Deel contract goes out the same day as the call and must be signed within 48 hours.
3. Start date.
4. The computer arrangement (see below).

Computer policy

Every hire starts on a clear review cadence, communicated at the offer and reinforced on day one.

• 30-day review — ramp progress, are they enjoying the work, are we enjoying working with them, is there anything visibly off-track.
• 90-day review — are they on the trajectory toward the 6-month success picture we wrote during Stage 0. This is the most important review of the first year.