This page, a companion to the HECC Manifesto, catalogues the roles Wave 5 is lifting into demand — at the top, the judgment layer where humans direct the machine; at the base, the physical and skilled trades the AI buildout depends on. Distinct from the Layer Split by Role, which takes a role you already hold and splits it into its two layers.
Every prior wave produced roles that did not exist before it. The plough produced the scribe. The factory produced the engineer. The internet produced the web developer. Wave 5 is producing its own.
This directory names them — plainly, without job-board filler — and shows what they actually do. Each entry describes the role in concrete terms and connects it to the part of the wave it sits in — the judgment layer at the top, or the physical and skilled trades at the base.
The list is selective, not exhaustive. The point is to make the elevated layer visible by pointing at the real jobs already appearing across the economy.
Judgment Layer Roles
Context architect
The role. A context architect knows a business well enough to tell AI exactly what to build — how to frame the problem, what a good answer looks like, where the model will quietly go wrong. It sounds technical. It isn’t. It’s judgment, pointed at a machine.
In practice. A context architect spends their week writing the briefs the AI works from, reviewing what it produces, and catching the confident-but-wrong outputs before they leave the building. They sit between the people who know the business and the system doing the work — translating one into the other. They are the reason an AI rollout produces useful output instead of polished noise.
Context architect is one shape the judgment layer takes in Wave 5. It is not a job title — it is a description of where the work is moving.
AI automation specialist
The role. An AI automation specialist walks into an organization, finds the cognitive work ready to compress, and builds the system that compresses it. They don’t do the execution. They decide what can be handed to the machine, and design the handoff.
In practice. An AI automation specialist spends their week mapping workflows, identifying which steps can be specified clearly enough for AI to run, and building the pipelines that connect tools, models, and human checkpoints. They sit between operations and engineering — close enough to the work to see what’s repetitive, close enough to the system to know what’s automatable. They are the reason a company stops talking about AI adoption and starts running on it.
AI automation specialists are the people who decide where the execution layer ends and the judgment layer begins — for everyone else in the building.
Forward Deployed Engineer
The role. An engineer embedded inside a client’s team to turn a frontier AI model into a system that actually runs in production. They don’t research the model. They make it useful in the world the client lives in.
In practice. A Forward Deployed Engineer spends their week inside customer environments — mapping where AI can plug into existing workflows, building the integration end-to-end, debugging the failures that only show up in real data, and feeding what they learn back to the model team. They sit between the lab and the customer — close enough to the model to know what it can do, close enough to the business to know what’s worth doing. They are the reason an AI rollout produces revenue instead of demos.
Forward Deployed Engineer is the role frontier labs invented when they realized the model wasn’t enough — and OpenAI, Anthropic, and Salesforce are spending billions to staff it.
AI Red Teamer
The role. Breaks AI systems on purpose. The job is to find the failures, manipulations, and edge cases before real adversaries do — and to document them clearly enough for engineers to fix. Adversarial creativity, applied to a machine that’s supposed to be safe.
In practice. An AI Red Teamer spends their week crafting adversarial prompts, attempting jailbreaks, testing where the model misbehaves under unusual inputs, and writing up findings detailed enough to ship as fixes. They work against both the model itself and the systems built on top of it — chatbots, agents, code copilots, anything an end user touches. They are the reason a model behaves the same way in front of an attacker as it does in front of a benign user.
AI Red Teaming cannot be automated because the whole point is creative human attack — finding what the system’s designers didn’t see. The work exists precisely because AI can’t yet be trusted to evaluate itself.
Evals Engineer
The role. Designs and runs the tests that decide whether an AI model is good enough to ship. Writes evaluation rubrics, builds regression suites, and flags when a new version breaks behavior that used to work. The work is judgment about what “good enough” actually means — codified.
In practice. An Evals Engineer spends their week defining what correctness looks like for a given task, building automated test suites that catch failure modes, running benchmarks across model versions, and surfacing the gaps between what the model claims and what it actually does. They sit between research and production — close enough to the model to know how it fails, close enough to the product to know which failures matter. They are the reason a model upgrade ships with confidence instead of with a prayer.
Evals are how the AI industry knows what its own systems can do. Every model release the world sees passed someone’s rubric first.
AI Auditor
The role. Verifies that AI models meet internal standards and external regulations. Reviews models for bias, accuracy, and compliance — and documents what they find in a paper trail regulators can read. The role exists because AI is moving into regulated industries faster than the rules can catch up.
In practice. An AI Auditor spends their week tracing where a model’s training data came from, testing outputs for biased or inaccurate behavior, documenting how the model performs under different conditions, and producing the compliance reports leadership and regulators need. They work in financial services, healthcare, hiring, and government — anywhere AI decisions affect people in ways the law cares about. They are the reason a regulated company can deploy AI without quietly breaking the rules.
As AI moves into regulated industries, someone has to certify that the model isn’t quietly breaking the law. AI Auditors are the people who sign their name to that certification.
Big Data Specialist
The role. Sits on top of mountains of data and decides what it’s actually saying. The execution — querying, cleaning, modeling — is increasingly handled by AI. What survives is the judgment of which questions matter and which patterns are signal versus noise.
In practice. A Big Data Specialist spends their week framing the questions the business actually needs answered, choosing which datasets are worth trusting, and translating raw patterns into decisions someone can act on. They sit between leadership and the data infrastructure — close enough to the business to know what’s at stake, close enough to the data to know when it’s lying. They are the reason a company stops drowning in dashboards and starts moving on what the numbers actually show.
Big Data Specialists turn information into judgment — the layer Wave 3 left scarce and Wave 5 lifts into higher demand.
Security Management Specialist
The role. Models threats, prioritizes defenses, and reads signals AI can’t yet weight. AI handles detection at scale; the specialist decides what’s worth defending, where the real risk lives, and which trade-offs are acceptable. Judgment, applied to an adversarial system.
In practice. A Security Management Specialist spends their week mapping where the organization is exposed, deciding which alerts are real threats versus noise, and choosing where to spend a finite defense budget. They sit between the executive team and the security tooling — close enough to the business to know what matters, close enough to the system to know how it breaks. They are the reason a breach doesn’t become a crisis.
Security Management Specialists prove the judgment layer isn’t optional — every business that runs on AI now runs on decisions only humans can be trusted to make.
AI / Machine Learning Specialist
The role. Decides what to build with AI, which models to deploy, and whether the output is trustworthy enough to ship. The engineering side of the title is real, but it’s increasingly compressed by AI itself. What stays human is the judgment about what’s worth building and what counts as good.
In practice. An AI/ML Specialist spends their week scoping problems worth solving with AI, choosing models and data, and stress-testing outputs against the edge cases that matter. They sit between research and product — close enough to the model to know its limits, close enough to the business to know which limits are acceptable. They are the reason an AI feature works in production, not just in a demo.
AI/ML Specialists build the systems doing the compressing. The judgment they bring — what to build, when to ship, what’s good enough — is the work that doesn’t compress.
Physical & Skilled Trade Roles
Data center electrician
The role. Wires the power systems the AI wave runs on. Every data center the buildout puts up needs more cable, more capacity, and more electricians than the local grid was designed for. The work cannot be automated because it cannot be specified clearly enough for a machine to perform.
In practice. A data center electrician spends their week pulling high-voltage cable, terminating connections to switchgear, and commissioning power systems that draw more than a small city. They work alongside cooling engineers and grid technicians, and the schedule runs in three shifts because the buildout is faster than the workforce can keep up with. They are the reason a billion-dollar data center can actually be switched on.
The companies spending billions to compress your cognitive work cannot move a dollar of that money without the electrician who pulls the cable.
HVAC / cooling systems engineer
The role. Designs and runs the cooling systems that keep AI infrastructure from melting itself. The chips at the heart of the AI wave run hot enough that cooling has become the single largest physical constraint on the buildout — bigger than power, bigger than space. Engineering-grade trade work that AI cannot do.
In practice. An HVAC/cooling engineer spends their week designing chilled-water loops, sizing pumps and heat exchangers, and commissioning cooling plants the size of small power stations. They work in spaces measured in megawatts of heat rejected per hour, and the design decisions they make today commit the building to its operating costs for the next decade. They are the reason the GPUs don’t shut themselves down.
The AI wave runs on judgment at the top — and on cooling at the bottom. Both are work that cannot be handed to the machine.
Structured cabling technician
The role. Connects the network — physically. Every server, every switch, every storage rack in a data center is wired together by hand, in copper and fiber, to tolerances no machine can hit at scale. The AI wave needs networks at a scale the trade has never built before.
In practice. A structured cabling technician spends their week pulling and terminating thousands of copper and fiber connections, testing each one with a certifier, and dressing the cabling into the racks so it can be traced, maintained, and replaced years later. They work alongside electricians on power and HVAC engineers on cooling, on builds where a single data hall holds tens of thousands of links. They are the reason the AI infrastructure can talk to itself.
A data center is a cathedral of cable. The judgment layer at the top runs on the cable at the bottom — and the cable doesn’t run itself.
Chip fabrication technician
The role. Operates the equipment that produces the chips the AI wave runs on. Semiconductor fabs are among the most complex industrial facilities on earth, and the people who run them work to tolerances measured in nanometers. The work cannot be remote and cannot be automated end-to-end — the precision lives in the hands.
In practice. A chip fab technician spends their week running lithography and etch tools, calibrating to spec, and troubleshooting when a wafer comes out of process. They work in cleanrooms in protective suits, on shifts that run 24/7 because a fab that idles costs millions a day. They are the reason the silicon the AI wave runs on actually exists.
Every AI model the world uses runs on physical hardware someone in a cleanroom had to build. The chip is the foundation the whole wave runs on.
Grid / transformer specialist
The role. Rebuilds the electrical grid to carry the load the AI wave is creating. The infrastructure that powered the last century cannot carry what the next decade demands, and someone has to physically install, upgrade, and maintain the transformers and substations that bridge the gap. A transformer doesn’t install itself.
In practice. A grid specialist spends their week siting and installing high-voltage transformers, upgrading substations, and coordinating with utilities and data center operators on power delivery schedules. They work on infrastructure that takes years to build and decades to outlive, and lead times on the equipment itself can stretch past the patience of the companies waiting for it. They are the reason the lights stay on as the AI buildout scales.
The grid is the constraint nobody talks about — and the specialists rebuilding it are looking at a decade of demand the rest of the economy cannot supply.
Crane operator / concrete crew lead
The role. Builds the physical shell the AI wave lives inside. Before a data center can be wired, cooled, or powered, it has to be poured, framed, and lifted into place. The work is older than the wave but the demand is being driven by it — the same companies building AI are building warehouses by the hundreds.
In practice. A crane operator or concrete crew lead spends their week placing structural members, pouring foundations rated for hundreds of tons of equipment, and coordinating with the trades who follow. They work on schedules driven by tech-company timelines, on sites that didn’t exist a year ago, and the pay reflects how scarce experienced operators are. They are the reason there’s a building at all.
The AI wave runs on buildings that didn’t exist five years ago. The people who pour them and lift them into place are walking into a construction boom that’s breaking records the industry has never seen before.