Skip to content
Recruiting Methodology

How we source the engineers who don't apply.

A senior AI/ML search runs in 8 to 10 weeks. Below is the IQC sourcing flow, the deliverables at each milestone, and the bar we hold every candidate against.

The IQC sourcing flow

Week 1

Calibration

90-minute call with the hiring manager. We co-author the IQC scorecard: must-haves, nice-to-haves, comp band, location, stage preferences. Client signs off. Search begins from a shared document, not an evolving conversation.

Week 2

Target mapping

A named map of 30 or more companies where the candidate currently sits. Specific functional groups within each company. Candidate-count estimates by domain. This is what makes the search a hunt, not a fishing expedition.

Weeks 2 to 6

Active outreach

Targeted approach to 30 to 80 passive candidates per role. Multi-channel: LinkedIn, email, mutual connections, conference rosters, GitHub. Each candidate gets bespoke outreach, not a template.

By end of week 6

IQC delivery

4 to 6 candidates delivered, each one meeting all six IQC criteria. Each comes with a written conviction memo: why they fit, what concerns to probe, comp expectation, references already verified.

Weeks 6 to 10

Interview process

Tier A: you run the process. Tier B: we manage scheduling, debriefs, offer construction, negotiation, resignation support, and 30/60/90-day check-ins.

What is an IQC?

A candidate is an Interview Qualified Candidate when all six are true:

  1. We have spoken to them directly, not just sourced them from a database.
  2. Their verified skills meet every must-have on the agreed role scorecard.
  3. They have expressed specific interest in this role at this company.
  4. Their base-comp expectation falls within the band you provided at calibration.
  5. They have confirmed availability and willingness to commit to your interview process.
  6. We have completed a light pre-screen: one reference signal or basis-of-employment verification.

This bar is materially higher than what most contingent firms send. It produces fewer candidates per dollar spent than a typical recruiter pipeline. That's the point.

Sample artifacts (anonymized)

We will publish anonymized excerpts of a target company map, a role scorecard, and a candidate conviction memo here once the methodology pack is finalized.

Methodology pack publication target: Q3 2026.

Start with the Calibration Sprint

Book the Calibration Sprint