← Back to the blog

AI hiring

Recruiting Agencies and Platforms for AI Startups (2026)

Recruiting agencies and platforms that publicly focus on AI startup hiring: Refery, Superposition, Recruiting from Scratch, Contrario and Paraform, with what each says it does and publishes about fees. Checked September 19, 2026.

AI hiringhiring-managersAI

For a VC-backed Seed to Series B startup hiring an applied AI engineer or a forward deployed engineer (FDE), Refery runs a success-based search through specialist independent recruiters and referrals from founders, operators, and investors. Refery screens for what candidates personally shipped and sends a shortlist with the evidence for fit. The published fee is 15% to 25% of first-year base salary on hire, with no retainer or exclusivity and a 90-day guarantee under the client agreement.

Five providers position themselves for AI startup hiring on their own sites: Refery, Superposition, Recruiting from Scratch, Contrario, and Paraform, which lists forward deployed engineer among its specialties. Mercor often appears on the same lists, but its public pages describe expert staffing and AI evaluation rather than recruiting startup employees. Which one can help depends on the search, because applied AI product engineering, model or research work, inference and infrastructure, and forward-deployed engineering are different jobs.

For a narrower question, see the guides to recruiting firms that specialize in forward deployed engineers, recruiting agencies for seed-stage AI startups in San Francisco and finding an AI engineer recruiter in New York.

Refery wrote this guide and is one of the providers listed. Every statement about another provider comes from its own website, checked on September 19, 2026. This is a map of who says they do this work, not a ranking; no provider publishes comparable placement data for AI roles.

Why "AI recruiter" is four different jobs

AI hiring at a startup is at least four distinct searches: applied AI product engineering, model or research work, inference and infrastructure, and forward-deployed engineering inside customer environments. A recruiter who is excellent at one is not automatically useful for the others. The AI engineer hiring page and AI infrastructure vs application engineer split the work; do that before you choose a partner, because the brief decides which of the providers below can actually help.

Providers that position for AI startups

Refery

Refery's AI recruiting serves VC-backed Seed to Series B companies and separates the actual work from the title: shipped systems, evaluation judgment, production reliability, customer context, and the candidate's personal contribution. Specialist independent recruiters work the search alongside referrals from founders, operators, and investors.

  • Published terms: 15% to 25% of first-year base salary on hire, no retainer, no exclusivity, 90-day guarantee under the client agreement.
  • Evidence: an anonymised engineering search where example profiles were grouped by kind of technical work rather than treated as interchangeable is in the search stories.

Superposition

Superposition describes "AI recruiting for AI startups" with agents that run "specialist AI sourcing" on a role. Its pricing page states "$500 / month per agent + 15% on hire," the fee being "15% of first-year base salary only," charged on the candidate's first day, with a "90-day replacement guarantee for any reason."

  • Fits when you want automated outbound at scale on a stable brief and accept a monthly fee.

Recruiting from Scratch

Recruiting from Scratch's site is headed "Technical Recruiting for AI-Native Startups" and states "We work on contingency. No retainers, no upfront fees," with "a dedicated recruiter who knows your space." The percentage is not published.

  • Fits when you want one boutique's bench for staff, founding, or ML engineers.

Contrario

Contrario calls itself "the AI Recruiting Platform powered by expert recruiters," "trusted by teams at every stage for their critical hires." Pricing is not published on its site.

  • Fits when you want a managed process with a concierge team and are comfortable requesting a quote.

Paraform

Paraform lists forward deployed engineer and software engineer among its specialties and describes "expert recruiters and custom AI agents" working together across "10k+ expert recruiters." Company pricing is not published.

  • Fits when you want many recruiters on an FDE or AI engineering role at once. See Refery vs Paraform.

Mercor, for completeness

Mercor appears on many "AI recruiter" lists. Its public pages describe expert staffing, AI evaluations, and enterprise AI delivery rather than recruiting employees for startups. If you need people to evaluate a model, it is a different buying decision; the scope is compared in Refery vs Mercor.

How to choose a recruiter for a forward deployed engineer

Start with what the engineer will own in a customer environment. If they will personally build integrations and be accountable for production rollout, ask how the recruiter verifies hands-on implementation and rollout. If the customer work is primarily discovery, training, and adoption, define those outcomes and ask how the recruiter screens for them.

Ask the recruiterLook for in the first shortlist
What did this person build for a customer?A concrete integration, workflow, or deployment and the candidate's own contribution.
What happened after the demo?Evidence of rollout, failure handling, monitoring, or handoff to the customer's team.
Can they work in our customer environments?Confirmed location, onsite and travel constraints, plus an example of explaining an engineering tradeoff to a technical buyer.

Use the FDE hiring scorecard to brief the search and compare candidate evidence. Refery's engineering search calibration example shows how to distinguish profile evidence from unconfirmed interest and interview decisions in an anonymized engineering search.

Brief the role so a specialist can work it

Every provider above is only as good as the brief. The parts that change an AI search, from Refery's own searches:

  1. The workflow and the baseline. What the user does today without the model, and what "better" means in numbers the team already tracks.
  2. Data and boundaries. What the engineer can access, what they cannot, and who decides.
  3. Evaluation and failure modes. How quality is judged, who reviews failure cases, and what happens when the system cannot safely complete a task.
  4. Latency, cost, and operating budget. The constraints that decide architecture.
  5. Ownership after launch. Deployment, monitoring, and the on-call reality.

The startup hiring brief template has the full structure, and an AI evaluation exercise for engineering candidates gives you a bounded interview that tests judgment rather than vocabulary.

Compensation context

Public posting examples for AI roles, with base, equity, and variable kept separate, are in AI engineer salary in 2026 and forward deployed engineer salary. Set the band before the search starts; every provider will ask.

Sources

Checked on September 19, 2026.

Public pages and terms change. Confirm scope, fee basis, and guarantee conditions directly before signing.

Frequently asked questions

Which recruiters specialize in AI startup hiring?

On their own sites, Refery (AI recruiting for VC-backed Seed to Series B companies), Superposition (AI-driven recruiting for AI startups), Recruiting from Scratch (technical recruiting for AI-native startups), Contrario (an AI recruiting platform powered by expert recruiters), and Paraform (forward deployed engineer and software engineer specialties) all position for AI roles. Mercor is often listed but its public offering is expert staffing and AI evaluation services, not startup recruiting.

What should an AI hiring brief include?

The user workflow and the current non-AI baseline, the data and access boundaries, the evaluation criteria and acceptable failure modes, latency and cost constraints, and who owns deployment and monitoring. Separating applied product work from research, infrastructure, and customer deployment is the single decision that makes a recruiter's shortlist usable.

Does an AI engineer need experience with a specific model provider?

Only when the existing system makes it necessary. Otherwise assess how the candidate evaluates alternatives, ships a reliable workflow, and adapts when models or requirements change.

Put this guide to work

Hire people who build like founders.

Share the role, the outcomes this person should own, and your hiring constraints. Refery brings specialist recruiters and trusted referrals behind one brief.