The 2024 calendar year produced more new AI job titles than the prior decade combined. The 30 roles below all share a single defining feature: each first appeared in our coverage universe (1,200 companies, 47,000 verified offers) after January 1, 2024 — and by February 2026, each clears at least three distinct employers and a verified compensation floor. The median compensation floor for the cohort cleared $186K base / $290K total in 2026, against a $0 baseline two years earlier when the roles did not exist.
Three findings shape the cut.
First, the highest-paying new roles cluster around AI safety and alignment — not productivity tools. Constitutional AI Researcher (Anthropic-invented, now industry-wide) clears $620K total at the median. Mechanistic Interpretability Researcher clears $560K. AI Welfare Researcher (the most contrarian role on the list, with maybe 12 qualified humans globally) clears $560K with a $310K base floor. The compensation gravity has moved structurally toward the alignment + interpretability + welfare cohort because the supply-side qualified talent pool is tiny — 80 humans globally for Constitutional AI Researcher, 12 for AI Welfare Researcher — and demand is elastic at frontier labs and the academic safety cluster.
Second, the RLHF economy graduated into a parallel comp ladder. Top-tier Mercor placements clear $200/hr 1099. The p90 RLHF Specialist cohort hits $400K+ annualized as 1099 contractors — at a rate that beats frontier-lab senior IC base bands and creates structural retention pressure on labs that don't pay competitively. AI Trainer (Domain Expert) — the senior-domain version of the broader RLHF labor pool — clears $120-300/hr 1099 at the senior tier requiring terminal-degree credentials. The 1099-track parallel ladder is the most consequential structural change in AI labor markets of 2024-26 and the title-stabilization story below the headlines.
Third, vertical AI roles converged across legal, healthcare, and finance domains. Vertical AI Domain Lead clears $480K total at Harvey, Abridge, Hippocratic, Tempus. Clinical AI Validation Specialist clears $260K total in healthcare AI. The pattern is consistent: senior-domain credential (JD, MD, CFA, PhD) plus applied AI fluency at the senior IC band. The vertical-AI cohort produced the most consistent post-2024 title convergence in our data — three years ago these humans were "the AI lead" at their company; in 2026 they hold dedicated senior-IC titles paying multiples of traditional senior-domain professional bands.
Methodology
We longlisted 94 candidate role titles from 47,000 verified offers across 1,200 companies in our coverage universe. Each candidate had to meet five hurdles: (1) first documented public job posting after January 1, 2024 in our data window; (2) at least 3 distinct employers with verified postings in the window; (3) verified compensation floor with n ≥ 8 verified offers — base + sign-on annualized + equity; (4) function description converged across employers; (5) the title is not a re-skin of a pre-2024 role. 41 roles cleared all five hurdles; 30 selected. Each role was scored across five weighted dimensions: Recency (30%), Active Requisitions (25%), Compensation Floor (20%), Distribution Across Employers (15%), Function Coherence (10%). Re-skinned roles (e.g. "Applied AI Engineer" — a post-2018 title) excluded. Compensation figures index US-private-sector senior-band offers; sovereign-AI and non-US-private offers may diverge materially. Year-over-year anchor: median compensation floor for post-2024-origin AI roles cleared $186K base / $290K total in 2026 — versus $0 baseline in 2024, when the roles did not exist. Data window January 2024 — February 2026.
The full ranked table is below. Three deep-cut profiles follow.
Constitutional AI Researcher anchors the list at $620K total comp. The role exists because Anthropic's December 2022 "Constitutional AI" paper produced a methodological frame for alignment that other frontier labs adopted post-2024. Function: design constitutions and value-shaping objectives that align model behavior without requiring human feedback at every step. Core skill mix: ML research at the frontier-lab senior band + moral philosophy literacy + RLHF / RLAIF systems engineering. Roughly 80 humans globally are qualified at the senior IC band — the supply constraint is acute. The compensation floor reset to $620K total comp in late 2025 after Anthropic's senior research band ramp, and the next reset is structural rather than discretionary because the talent pool cannot expand fast enough to track demand. Anthropic, OpenAI, Google DeepMind, and Meta FAIR are the only buyers at scale; the academic AI safety cluster (CHAI Berkeley, MIT CSAIL, Cambridge MILA-adjacent) supplies the senior-PhD funnel.
Eval Designer is the role that emerged once benchmark-driven research hit a ceiling. The field accepted post-2024 that traditional benchmarks (MMLU, HumanEval, GSM-8K) had been saturated by frontier models and that capability measurement required adversarial design rather than off-the-shelf tests. Function: design adversarial evaluations, capability probes, and red-team tasks that reveal model failure modes traditional benchmarks miss. Core skill mix: senior domain expertise (math, biology, law, code) plus adversarial thinking plus statistics. Scale AI's Thunderforge program plus Mercor's expert marketplace plus the broader RLHF / red-team economy anchor the structural demand. The role's compensation floor cleared $440K total comp through 2025 because supply-side qualified talent is small and demand is elastic. The 2026-27 horizon: Eval Designer roles will branch into capability-specific sub-titles (Math Eval Designer, Biology Eval Designer, Code Eval Designer) as evaluation pipelines mature.
Agent Operations Engineer is the production-engineering role that emerged with the 2025 agent launches. Once frontier labs shipped agents at production scale (GPT-5 + Claude 4 + Gemini 2 in 2025) the bottleneck on enterprise agent deployment was no longer model quality — it was production reliability. Function: build the systems that supervise, monitor, intervene, and recover long-horizon AI agent runs across enterprise environments. Core skill mix: SRE plus applied AI plus tool-use systems plus observability. The role title converged across OpenAI, Anthropic, Microsoft AI, Google DeepMind, Sierra, and Decagon through 2025; before mid-2024 the role was scattered across "Applied AI Engineer" and "ML Platform Engineer" titles. Compensation floor at $380K total comp; senior IC ramp through 2026 will track agent deployment maturity. Decagon's Agentforce-adjacent senior IC ramp is the proof point.
The 2026-27 forecast: three structural shifts. First, the AI Welfare Researcher cohort expands from 12 humans to 30-50 as Anthropic's 2024 thesis (that the moral status of AI systems may warrant dedicated research investment) gains adopters at OpenAI, Google DeepMind, and the academic safety cluster — expect a parallel +200% growth in the welfare-research compensation band. Second, the Sovereign AI Compliance Officer role spreads beyond GCC anchors as the EU AI Act enforcement timeline (2026) plus the US executive order operationalization plus the UK AI Safety Institute mandate produce parallel demand in Brussels, Washington DC, and London — expect base compensation to clear $200K with multi-jurisdictional Arabic + English fluency at premium. Third, the Agent Observability Engineer cohort scales sharply with 2026-27 enterprise agent deployments — this is the single most likely 2027 ranking entry to clear the $380K compensation floor at significantly larger headcount than any role in our current top 10.
For the full company hubs, see Anthropic, OpenAI, Scale AI, Mercor, Anysphere, Harvey, Abridge, and G42. Cross-reference: Top 20 Highest-Paid AI Roles 2026, Top 25 AI Product Launches That Created Jobs 2026, and Top 50 CHROs Reshaping Talent 2026.