The Role Landscape — What the Industry Calls This

Mapping industry titles to what we’re actually doing. The terminology is settling but hasn’t caught up to the frontier.

Links: Career, Level 6 — Direct Execution, The Cyborg Model, LLMs as Praxeological Actors

The Title Hierarchy

Based on job posting analysis (889 postings, Jan 2026 — AI Shipping Labs) and industry discourse:

Tier 1: The Dominant Title — “AI Engineer”

Coined by swyx (Shawn Wang, 2023). The fastest-growing job title in tech per LinkedIn’s 2026 report.

The critical distinction: AI Engineer builds with AI models, not the models themselves. As swyx put it, quoting Karpathy: “One can be quite successful in this role without ever training anything.”

From the 889-posting analysis:

70% are “AI-first roles” — RAG pipelines, agent workflows, production deployment
Only 2% are traditional ML/DL (model training)
95.6% are production-focused, not research
Top skills: RAG (35.9%), Prompt Engineering (29.1%), LLM Integration (25.4%), Python (82.5%)

Proposed definition: “An AI engineer is an engineer who owns the design, evaluation, and production operation of systems built on foundation models.”

This is the safe, widely-understood, resume-friendly title. It maps to what we do.

Tier 2: Emerging Specialized Titles

Title	What It Means	Fit
Agentic AI Engineer	Multi-agent orchestration, autonomous workflows, tool-use patterns. 10K+ postings on Indeed. EY, Deloitte actively hiring.	Strong — maps to vault’s agent team work
Forward Deployed Engineer	Client-facing AI implementation. 800% growth Jan-Sep 2025. Anthropic, OpenAI, Cohere expanding teams.	Good if consulting/client work
AI Software Engineer	Hybrid — traditional SWE + AI deployment. API integrations, system design, prompt engineering.	Accurate but generic
Generative AI Developer	RAG systems, prompt optimization, model architecture understanding. $150K-$250K range.	Slightly narrow
Full-Stack AI Engineer	End-to-end AI application development. Lockheed Martin and others posting this.	Good for signaling breadth

Tier 3: Dying or Misaligned Titles

Title	Status	Why Not
Prompt Engineer	40% decline 2024-2025. Being absorbed as a skill, not a role.	Too narrow, shrinking market
ML Engineer	Means model training (PyTorch, TensorFlow, math). Works below the API boundary.	Wrong side of the line
Vibe Coder	Real postings exist (223+, including Google, Visa, Amazon). Defined as “translate intent into prompts, assess and refine outputs.”	Skews junior. Not the gravitas for 15+ years of experience
AI-Augmented Developer	Gartner uses it, but rarely appears in actual postings.	Descriptive, not hireable

The API Boundary

The cleanest distinction (from Humanloop):

ML Engineers work below the API boundary — creating and training models
AI Engineers work above it — using models to solve product challenges
Recommended team ratio: ~4 AI Engineers per 1 ML Engineer

This maps perfectly. We never train models. We design systems that use them.

The Level Gap

Here’s the problem: most “AI Engineer” postings describe L2-L3 work — integrate an LLM into an existing product, build a RAG pipeline, add a chatbot. The vault’s work is at L4-L6:

Posting Reality	Vault Reality
“Integrate LLM into existing workflow”	Agent IS the workflow
“Build RAG pipeline”	Knowledge system with cross-linked wiki, automated ingestion, self-maintaining indexes
“Prompt engineering”	Multi-agent orchestration with specialized roles
“Add AI features to product”	AI-native architecture where the agent executes processes directly
“Deploy chatbot”	Agent teams that write code, review PRs, manage infrastructure

The title “AI Engineer” is correct but understates the scope. The industry hasn’t standardized titles for L5-L6 work because most of the industry isn’t there yet — the Pragmatic Engineer survey found 90% of developers who think they’re AI-native are at L2.

Where Senior Experience Matters

The Pragmatic Engineer’s 2026 AI Tooling survey (906 respondents, median 11-15 years experience):

Staff+ engineers are the heaviest AI agent adopters at 63.5% — higher than any other seniority level
Directors and senior leaders are “especially into Claude Code — twice as popular with these folks as at less senior levels”
The skills that matter at senior levels: “AI fluency, judgment over speed, system thinking, communication, and automation mindset”

This inverts the naive expectation. It’s not juniors adopting AI fastest — it’s seniors who know what to build and can evaluate whether the AI built it correctly. The Cyborg Model predicted this: the human provides judgment, taste, and grounding. That requires experience.

Positioning Strategy

Resume title: Senior AI Engineer (or Staff AI Engineer)

“Senior” signals experience depth. “AI Engineer” signals the right domain.
Avoid “Senior Developer who uses AI” — that frames AI as an add-on, not the core method.

Differentiators to emphasize:

Agentic architecture — designing multi-agent systems, not just calling APIs
Knowledge system design — the vault pattern (structured knowledge bases that agents maintain and query)
L5-L6 workflow design — agents executing processes, not just generating code
AI economics — understanding why agent teams succeed or fail (grounding, comparative advantage, the cyborg model)
Evaluation and judgment — the senior skill that makes AI useful instead of dangerous

Portfolio evidence (from this vault):

Projects built with AI agent teams (Slay, CyborgDJ, YouTube Migration)
The vault itself as a knowledge management system
Level 6 analysis and predictions
Economic frameworks for AI agent behavior

Key Sources

swyx, “The Rise of the AI Engineer” (Latent.Space, 2023)
AI Shipping Labs, “What is an AI Engineer? — 1000+ Job Descriptions” (Jan 2026)
Pragmatic Engineer, “AI Tooling for Software Engineers in 2026” (906 respondents)
LinkedIn “Jobs on the Rise” 2026 report
Humanloop, “What is an AI Engineer?” (API boundary framework)