What team and skills does a first quantum project need?

Who gets hired first — and what does the record say?

The first profile a first quantum project needs is a classical optimization operator, not a quantum physicist — because without one, the project has no number to beat. Rosetta Quantum has watched this casting error from both sides of its own bench. Every verdict in our ledger is a comparison against a classical baseline, and the profile that produced that baseline is a solver operator running open-source classical software: in our entry 001, OR-Tools CP-SAT reached the exact optimum of a 12-asset portfolio instance in 0.113 s while the quantum arm (QAOA, simulated) landed 42.8% away at equal budget (our run-001, sealed 2026-07-23). The person who set that bar wrote zero quantum code.

The pattern generalizes beyond our bench. Our pilot-evaluation ladder orders the evidence gates cheapest-kill-first: Gate 0 is the tuned classical incumbent on your real instances (license price of the number-to-beat: $0 — OR-Tools and CP-SAT are Apache-2.0, per Google's own repository, accessed 2026-09-12); Gate 1 is the candidate circuit on a noise-free simulator; only Gate 2 touches paid hardware. A weak classical baseline invalidates everything downstream — that is the most common way a quantum benchmark dies on review. Staff Gate 0 badly and the rest of the project is theater.

What does the hiring record actually measure?

It measures that most of what the quantum industry hires is not quantum-specific — which is good news for a first project, because it means most of the team is probably already on your payroll. The public paper trail, dated:

The Chicago Quantum Exchange analyzed 5,187 accessible job postings (of 10,484 collected, 2018–2023, drawn from QED-C and Quantum Computing Report job boards) and found that in 2022–23, 55% of quantum-industry postings required at most a bachelor's degree — 14% asked for a master's and 31% for a PhD (CQE study, published 2024-09-10). In the industry segment specifically (excluding academia and national labs), roughly two-thirds of positions were open to bachelor's-or-less profiles.

That posting-level picture matches what companies say when asked directly. A QED-C survey of 57 member companies found the industry hires "a range of job opportunities" from narrow specialist roles (quantum algorithm developer, error-correction scientist) to broad business, software and hardware roles — and that the broad roles "require a range of skills, most of which are not quantum related" (Hughes et al., arXiv:2109.03601, survey run 2021, published IEEE Transactions on Education 2022).

The scarcity is real but narrow. McKinsey estimated roughly one qualified quantum candidate per three quantum job openings and projected that fewer than half of quantum jobs would be filled by 2025 without intervention (McKinsey, 2022-12-01 — a dated estimate and a dated projection; we found no 2026 re-measurement of that ratio). Meanwhile the money side grew an order of magnitude: $12.6B of private investment went into quantum technology startups in 2025 (6.3× 2024) and about one in three large companies surveyed allocated $10M+ to quantum initiatives in 2025 (McKinsey Quantum Technology Monitor 2026). Scarce specialists plus growing budgets is exactly the environment where a first project overpays for the wrong profile first.

Which profile does each evidence gate need?

Three profiles you almost certainly already employ, plus one you rent later. Mapping the team to our evidence gates:

The division of labor matters more than the org chart. Gate 0 produces the denominator of every later claim; the profile that runs it needs to be good at classical optimization, benchmark hygiene and instance realism — a skill set that predates quantum computing and is hired on the general market. Gate 1 is software engineering: installing an SDK, building the circuit the paper describes, running it noise-free. The quantum-specific judgment — which mitigation to pay for, which qubit topology fits, how to read a vendor's native-gate pricing — first binds at Gate 2, and at first-project scale it is a consulting engagement or a fractional hire, not a permanent line on the payroll.

When does a first project actually need a quantum specialist?

At Gate 2 — and renting the judgment is usually cheaper than owning it. The specialist's first real contribution is not writing the circuit (Gate 1 staff did that from public docs); it is keeping the hardware run honest: shot budgets, error-mitigation configuration (which can multiply run cost — we published the mechanics in our error-mitigation explainer), transpilation choices, and reading results without over-claiming. That last part is the one Rosetta cares about: a team whose only quantum expertise arrived with a mandate to find quantum advantage has an incentive problem — the profile most likely to kill the pilot cheaply (the classical operator) reports to the profile whose job depends on the pilot surviving. Keep the classical arm independent; we wrote the failure mode up in our benchmark-cheating post.

What a first project does not need on day one, per the record above: a quantum physics PhD on payroll (31% of postings ask for one, and those postings cluster in hardware vendors, academia and national labs — CQE 2024, 2022–23 data), a proprietary algorithm, or hardware commitments before Gate 1 produces a best-case number. If the noise-free best case cannot approach the Gate-0 baseline, hardware cannot save it — and no hire fixes that either.

What we don't know

• Postings are not hires. The CQE dataset measures what companies advertised, not whom they hired or retained; degree requirements listed in a posting are a floor, not a skills inventory.
• The freshest posting data ends in 2023. The CQE analysis was published 2024-09-10 on 2018–2023 data; we found no equivalent 2025–26 postings dataset of comparable size. The degree mix may have shifted.
• The 1:3 candidate ratio is a 2022 estimate. McKinsey published it 2022-12-01 and we found no re-measurement since; training pipelines have grown (55,000+ students enrolled in one national quantum course initiative as of February 2026, per McKinsey QTM 2026), so the ratio has plausibly moved.
• The data is US-centric. CQE's job boards and QED-C's membership skew American; degree culture differs by market.
• Our own evidence is N-of-our-bench. Rosetta's claim that the classical operator produced every baseline number is a fact about our ledger, not a law about your project.

Sources

• CQE — Quantum Tech Job Accessibility study (2024)
• CQE news — most quantum jobs don't require a graduate degree (2024-09-10)
• Hughes et al., Assessing the Needs of the Quantum Industry — arXiv:2109.03601 (IEEE Trans. Educ. 2022)
• McKinsey — Five lessons from AI on closing quantum's talent gap (2022-12-01)
• McKinsey Quantum Technology Monitor 2026
• Google OR-Tools (Apache-2.0)
• Qiskit documentation
• Rosetta — How to evaluate a quantum pilot without burning budget
• Rosetta — Why a weak classical baseline ruins a quantum benchmark
• Rosetta — run-001: QAOA vs CP-SAT (sealed 2026-07-23)
• Rosetta — What is error mitigation and why does it make runs more expensive?
• Rosetta — How do you cheat on a quantum benchmark (and how to avoid it)?
• Rosetta — How to choose a quantum cloud provider without getting locked in

Rosetta Quantum publishes verdicts with raw, reproducible data. This is educational content, not a product claim.