The easy hires are behind you.
By Series B you have a recruiting function and it works. What reaches us at this stage are the roles it cannot close: the genuinely rare profile, the confidential backfill, the new site in a market nobody on the team knows.
What actually makes hiring hard here.
Your in-house team is good at the roles it sees repeatedly. It is structurally bad at three things: profiles where the credible pool is a few hundred people worldwide, searches that cannot be run openly, and hiring in a geography where the team has no network. Those are exactly the roles that matter most at this stage, and they are the ones that sit open for months.
Who to hire, and in what order.
Production ML infrastructure, perception and SLAM, low-level systems, specific security domains. The pool in any market is measured in hundreds, and most of them are in seats they like.
Replacing someone who does not know they are being replaced. Cannot be posted, cannot be broadcast to a wide network — routed to a named subset under NDA.
Ten to thirty engineers on one bar, in a market where nobody on your team has a network. Leadership hire first, then the pod, then a clean handover to your own TA team.
Running a confidential search through an open channel
Briefing a wide recruiter network on a confidential backfill means your team hears about it. Ask any partner specifically how confidential mandates are routed before you brief them — and if the answer is vague, that is your answer.
Frequently asked questions
The constraint is that the incumbent, the team and the market must not find out. That rules out posting the role and rules out broadcasting the brief to a large recruiter network. A confidential search goes to a small named subset of recruiters under NDA, with candidates told the company name only once they have signed one themselves, and interviews scheduled off-site or outside working hours. At 100Networks, confidential and leadership briefs go to a restricted named group rather than the full network.
Usually reach rather than capability. An in-house team is excellent at roles it sees repeatedly, because it has built a pipeline for them. It is structurally disadvantaged on profiles where the credible pool is a few hundred people, on searches that cannot be run openly, and in geographies where it has no network. Those are not skill gaps; they are network gaps, and they are the specific thing an external specialist is buying you.
Retained buys dedicated attention and a consultant who works the search whether or not it is easy, which matters for board-level roles and genuinely rare profiles. You pay in instalments regardless of outcome, typically 25–35% of first-year compensation. Contingency and marketplace models cost less and carry no risk if nobody joins. For most Series B and C engineering and leadership roles, contingency or a specialist marketplace is proportionate; reserve retained for board-reporting searches.
Leadership first, always. Hire the person who will run it before the team they will run — they set the bar, interview everyone after them, and are the reason good local candidates say yes. Then the core pod against one rubric, then a handover to your own TA function. Trying to hire the team first and the leader afterwards is the most expensive sequencing mistake in site build-outs, and it usually costs six to nine months.
Make hiring the thing you’re good at.
Seed, Series A to D, or enterprise — send us one hard role. We'll match it the same day, and you'll meet a shortlist in days. Pay only when they join.