A résumé tells you where someone has been. It has always been a weak proxy for what they can do, but in 2026 it is weaker than ever — because AI assistance has made the artifacts of competence (polished portfolios, articulate cover letters, even take-home submissions) cheap to produce.

The industry has noticed. Skills-based hiring — evaluating candidates on demonstrated ability rather than pedigree — has moved from HR-conference talking point to standard practice at roughly half of IT organizations. We built Markosh’s staffing arm on that premise from day one.

What engineer-led vetting actually means

Every candidate who enters the Markosh network is screened by a working engineer, not a recruiter with a keyword checklist. The screening is structured around three questions:

Can they do the work? Live technical evaluation on realistic problems — system design conversations, code review of real-world-shaped code, debugging under discussion. Not puzzle trivia, and not take-homes that measure free evenings more than skill.

Can they do the work with AI? This is the newest filter and increasingly the most predictive one. We evaluate how candidates direct AI tooling: whether they can decompose a problem, verify generated output, and catch the confident nonsense. An engineer who treats AI output as finished work fails this screen; an engineer who treats it as a fast first draft from a junior colleague tends to excel everywhere else too.

Can they work inside a client’s team? Communication in written form, expectation-setting, and the discipline of weekly demos. Embedded engineers succeed or fail on this more often than on raw technical depth.

Why the funnel is narrow on purpose

Most applicants do not pass. That is the point. A staffing bench is only as valuable as the confidence a client can place in the next person who comes off it. Wide funnels optimize for placement volume; narrow funnels optimize for the thing clients actually buy — not needing to re-interview our work.

What to take from this if you are hiring

Even if you never work with us, steal the model:

  • Weight demonstrated skill over credentials, especially for AI-era roles where the credential lag is worst.
  • Put a practicing engineer in the loop for every technical hire.
  • Add an explicit “works well with AI tooling” evaluation. It is the highest-signal screen we have added in years.
  • Treat communication as a technical skill and test it like one.

The gap between “looks qualified” and “is qualified” has never been wider. Vetting is how you close it.