Specialist-model research desk
Understand the capabilities behind the roles, responsibilities and pay ranges.
Full employer listing; San Francisco
$270,000–$300,000; the listing mixes “base salary” wording with “OTE”, plus equity/benefitsA need to connect customer data, experiments and compute to production outcomes
Full employer listing; San Francisco, Seattle or New York
$165,600–$207,000 stated US base salary; equity/benefits separateDedicated benchmark design and failure diagnosis, rather than training alone
Employer-indexed listing; hybrid; exact location not captured in this pass
$165,000–$330,000 labelled compensation, plus equityProduct and infrastructure work around how users train models
Employer-indexed listing; location and full terms not verified
$200,000–$260,000 labelled compensationCustomer-facing engineering and technical adoption capacity
Do not average these figures. Roles, levels, compensation definitions and evidence quality differ; listings may have closed.
Hiring can reveal the capabilities a company wants to build; it does not prove customer demand, a committed budget or an immediate need for an external supplier.
Sources: Together vacancy, Scale vacancy, Baseten vacancy and Fireworks vacancy.
The Together listing explicitly combines data strategy, algorithm choice, controlled experiments, task-specific evaluation and compute sizing.
Treat a role as one observation.
This is a make-or-buy decision about capabilities, not a contest between salary and a consultancy’s headline quote.
Use sanitised examples and a time-bounded exercise; do not solicit unpaid production work.
Use it only as evidence that these advertised roles carry material labour commitments.
Build a capability coverage sheet for your proposed pilot.
Short answers to the questions buyers ask.
Read →10 / REFERENCEEvidence gaps and limits that still need resolving.
Read →11 / REFERENCEThe research’s own quality scores and limitations.
Read →12 / REFERENCETrace the section’s claims to the original sources.
Read →Useful responsibility and recruitment evidence with explicit compensation caveats. Limited sample and uneven page access prevent broad hiring-demand or salary conclusions.