New AI Roles · Explore

Engineering has named its AI Builders. Business functions haven't.

Nuvepro is naming the layer ahead of the market — so you can hire for it, build readiness for it, and measure it.

Sourced from 179,570 real job postings · 1,488 AI-Builder postings · 1,107 distinct titles

11%
of all Anthropic openings

47 of 422 current Anthropic openings are Applied AI Architect or Forward Deployed Engineer variants. The frontier-AI labs are naming this layer first.

anthropic.com careers, May 2026
Levie
Box CEO, May 2026

“Akin to a forward deployed engineer for internal functions. This is why, at Box, we're starting to hire for AI automation engineering roles.”

x.com/levie
19x
more FDE postings in Jan 2026 vs Jan 2025

Indeed-tracked growth in Forward Deployed Engineer postings. Palantir-originated, now templated at OpenAI, Anthropic, AWS, Google. The same role is now pointing inward — see AI Accelerator in the engineering row.

Indeed via Business Insider, May 2026
The naming gap
7848
Engineering: 784 distinct titles across the market converge to ~8 canonical names. The market has named this layer.
228 → 0
Business functions: 228 distinct titles across Marketing, Finance, HR, Legal, Sales, Ops, CX → zero canonical names. Every posting invents the role from scratch. The work is happening; the title hasn't caught up.
Off the org chart

The AI trainer gig layer

Beneath every named AI role sits a contingent-labor layer: domain experts hired per-task to train, evaluate, and red-team models. Not employees, not headcount — a labor market parallel to the org chart. Scale AI's Outlier and Mercor are the largest platforms. Mercor reports ~30,000 contractors at an average billed rate near $95/hr; creative-writing trainers run $75-$150/hr. This layer shows up in /explore/new-roles as a footnote because it doesn't sit on the org chart — but it's where the judgment, taste, and domain expertise that tunes today's models actually comes from.

Sources: Mercor, TIME (2026), Business Insider (May 2026)

Build your AI Builders, name your layer.

Audit how many of these roles your org already needs. Run the Bootcamp to ship the first one in 14 days.

Methodology

Counts are computed from 179,570 job postings in Nuvepro's real-world JD corpus (jobscraper.db, snapshot 2026-05-18). Engineering buckets match by canonical title pattern. Business-function buckets match by function keyword AND any AI-flavor term (AI, GenAI, agentic, LLM, ML, prompt, applied AI). Operator buckets match canonical executive titles ("VP AI", "Head of AI", etc.) — strict to avoid over-counting.

"AI Builder" as a literal job title returns zero matches across the corpus. The category is Nuvepro's term; the market hasn't yet named it. Cross-row title overlap is removed by DISTINCT job ID at the row level.