A 14-Day Fill on a Role Two Competing Vendors Had Been Working for Four Months

The requisition had been open for one hundred and twenty-one days when the engagement reached Cloudhire.

A Series C B2B SaaS company, venture-backed, post-Series-C, scaling from forty engineers toward seventy over an eighteen-month plan, needed a Senior Backend Engineer with specific depth in distributed systems at high transaction volume. The technical bar was real but not exotic: Go or Rust in production, prior experience operating systems handling fifty thousand or more requests per second, hands-on familiarity with the consistency-and-availability tradeoffs that show up at that scale. The cultural bar was the harder one for a small senior engineering team, high velocity, no patience for candidates who needed weeks to ramp on a first non-trivial system.

Two staffing vendors had been working the role since the start of the cycle. Between them, they had run the requisition for one hundred and twenty-one days. They had submitted, by the VP of Engineering's count, "somewhere between forty and fifty" candidates. He had stopped tracking precisely after the first sixty days.

Of those candidates, six had reached the on-site final round. None had been offered. The pattern, when the VP described it, was specific and exhausting. Candidates would interview well on system design at a conceptual level, most senior engineers can talk about CAP theorem and consistent hashing and then fall apart on the operational questions. What did you actually do when this system, which you claim you ran, had a partition. What was the on-call rotation? Show me the postmortem of the last incident you ran. The candidates who had genuinely operated the systems they claimed to have operated could answer. The candidates who had read about those systems on engineering blogs could not. The vendors had no way to tell the two apart at submission, so the filtering was happening on the client's calendar.

By the time Cloudhire was brought in, the engineering team had absorbed the missing capacity into the existing senior engineers' workload. Two of those engineers had stopped attending optional meetings. One had updated her LinkedIn headline to remove the company name. The role had moved from "open" to "actively damaging the team." The VP described his timeline as "yesterday."

The instinct in the staffing industry is to read a 121-day vacancy as evidence that the role is hard to fill. Sometimes it is. More often, it is evidence that the vetting is hard, and the sourcing is being asked to substitute for vetting work it cannot do.

A staffing vendor working a senior engineering role with no in-house assessment infrastructure has only two filters available: the resume and the recruiter screen. Both are conversational. Both are gameable. A candidate who has spent twenty minutes on Hacker News reading about the technologies in the job description can pass a recruiter screen. A candidate who has invested an afternoon refining their resume can match the keyword filter. Neither candidate can operate the systems the company actually needs to operate.

The vendor cannot run a real technical assessment because the vendor does not have one. The candidate's first encounter with rigorous technical evaluation is the client's own interview loop, which means every weak candidate consumes the client's senior engineering time before being filtered out. Forty candidates submitted, six on-sites, zero offers the math is the inevitable output of a system where filtering happens at the wrong layer.

Cloudhire's submitted candidates do not have this problem because they have been filtered before submission. The filtering is the Adaptive Skill Index, running inside the Integrity Engine, meaning every candidate's technical ability has been measured at calibrated difficulty, in a session where assistance was actively prevented, with a score that tells us specifically how the candidate operates against high-volume backend assessment items. The filtering is the soft-skills evaluation, meaning we know which candidates have the feedback metabolism and async-collaboration profile that matches a small senior team. The filtering is the verified employment record, meaning when a candidate claims to have operated a system at fifty thousand requests per second at a prior firm, that claim has been peer-attested by people who would know.

A submission from Cloudhire is not a candidate to be vetted. It is a candidate who has been vetted, presented to a client whose interview cycle can therefore focus on the things that actually require the client's judgment.

Day one, requisition intake. Cloudhire's account team mapped the role's specific requirements, Go or Rust in production, fifty-thousand-RPS minimum operational experience, small-team velocity expectations into the search filters that would be applied against the existing candidate network.

Day two, internal search. The filter returned twenty-three candidates globally with verified backend experience at the required scale, in the required language stack, with ASI scores in the band the role required. Of those twenty-three, eleven were on active engagements and unavailable. Of the remaining twelve, four had geographic or compensation constraints that did not match the requisition. Eight were available, qualified, and matched.

Days three and four, candidate confirmation cycle. Each of the eight was contacted, briefed on the role, and confirmed availability and interest. Six confirmed. Two declined one because the company stage was earlier than they preferred, one because they were in a late-stage interview at another opportunity.

Day five, submission. Six candidates were delivered to the client, each with a complete Cloud-ID profile attached. The submission package included verified employment history with peer attestations from prior roles, ASI scores broken down across foundational reasoning, applied implementation, and system reasoning, soft-skills profiles flagged against the team's specific context, and the operational-experience claims relevant to the role specifically called out and verified.

Day six, the VP of Engineering reviewed the six. He scheduled four interviews. By his own description on a follow-up call, this was the first time in the four-month cycle that he had read a submission package and felt confident scheduling the candidate without first running an additional internal screen.

Days seven through ten, technical interviews. The four candidates ran through the engineering team's standard system design and operational deep-dive sessions. Three of the four passed cleanly. The fourth was strong technically, but the team-fit signal in the soft-skills profile flagged at submission surfaced in the interview, and the team made the call not to advance. Cloudhire had flagged the risk; the client made the decision; no time was wasted.

Day eleven, behavioral interviews with the broader team for the three remaining candidates. Two emerged as strong choices. One, in the VP's words, "made the rest of the team taller in the conversation."

Day twelve, reference cycle. Because the candidate's Cloud-ID already contained verified peer attestations from his two most recent roles, the reference cycle was completed inside a single business day.

Day thirteen, offer extended.

Day fourteen, offer accepted and placement logged.

The candidate started two weeks after acceptance. By his fourth week, he had owned and shipped a non-trivial latency-reduction project that had been deferred for six weeks waiting for the role to fill. By his ninetieth day, he had been promoted into informal technical leadership of a two-person sub-team, with the VP describing the trajectory as "two years' worth of trust earned in three months."

The two senior engineers who had been carrying the missing capacity stopped doing so by the candidate's third week. One of them, the one who had removed the company name from her LinkedIn, has since added it back.

The client has subsequently brought three additional engineering roles to Cloudhire and is in active conversation about a fourth. Two of the three have been placed.

The headline number fourteen days against one hundred and twenty-one is the thing readers will remember. It is not the thing that caused the outcome.

The cause is that the verification and assessment work that the prior vendors had been outsourcing to the client's interview loop had, in Cloudhire's case, already happened. The candidate who eventually accepted the role had been ASI-scored on backend system reasoning months before this requisition existed. His employment history at high-scale firms had been peer-attested. His soft-skills profile had been mapped against the kinds of teams Cloudhire's clients typically present to. His identity was Cloud-ID anchored and his prior assessments were Integrity-Engine cleared. When the requisition arrived, none of that work had to be done from scratch. It had to be matched.

A four-month vacancy and a two-week fill are not measuring the same kind of work. The four-month vacancy is measuring how long it takes to source, screen, vet, and verify candidates reactively, requisition by requisition, without infrastructure. The two-week fill is measuring how long it takes to match an existing pre-verified network against a specific role.

These are different operations. They produce different outcomes. The same client can experience both the first with vendors operating at the surface, the second with a network where the precondition work has been done in advance.