{"id":78,"date":"2026-07-26T19:22:36","date_gmt":"2026-07-26T19:22:36","guid":{"rendered":"https:\/\/www.dobryakov.net\/blog\/78\/"},"modified":"2026-07-26T19:22:36","modified_gmt":"2026-07-26T19:22:36","slug":"ai-demand-from-vacancies","status":"publish","type":"post","link":"https:\/\/www.dobryakov.net\/blog\/78\/","title":{"rendered":"AI Job Demand in 2026: What Hiring Signals Show in Job Postings"},"content":{"rendered":"

Hype says \u201chire AI.\u201d Real job descriptions buy people who own production agents<\/strong>, evals<\/strong>, Applied AI<\/strong>, and partner delivery<\/strong>. Below is a method for reading vacancy clusters as an org chart under pressure, plus five role clusters from a parsed corpus of open roles.<\/p>\n

<\/p>\n

Market wrap-ups love aggregates: \u201cAI demand grew N%.\u201d For a CTO or Head of Engineering that percentage is almost useless: it does not say whom<\/em> companies buy or which pain<\/em> the hire is meant to close.<\/p>\n

I take a cluster of roles<\/strong> from one company \u2014 or one type of signal \u2014 and look at the org structure that shows through the text. Responsibilities in a JD are often more honest than the keynote: they already name what does not scale, what fails on customers, and what is missing before the next round or a listing.<\/p>\n

The corpus covers open vacancies at a frontier lab, enterprise pharma, industrial Series C, and an agent startup preparing for Series A. It is not \u201call of LinkedIn.\u201d It is a method on concrete JDs: where the pattern repeats across employers, that is demand; a single polished recruiter line proves nothing by itself.<\/p>\n

Method: the JD as an org chart under pressure<\/h2>\n

One vacancy is noise. Three roles side by side with overlapping ownership is a drawing.<\/p>\n

Questions I put to the text:<\/p>\n

    \n
  1. What broke or stopped scaling?<\/strong> Often stated outright in responsibilities: \u201ccustomer outcomes,\u201d \u201cregression across product surfaces,\u201d \u201cpartner-led pre-sales.\u201d<\/li>\n
  2. Who is being hired next door?<\/strong> Three Engineering Managers at once, or EM + Prompt\/Evals + Partner SA, is not \u201cwe opened another seat.\u201d It is a delivery-model change.<\/li>\n
  3. Where does success of the role live?<\/strong> \u201cTake automation from ~40% to 85\u201395%,\u201d \u201cevaluation harness,\u201d \u201cproduction AI assets\u201d \u2014 that is no longer \u201cexplore LLMs.\u201d<\/li>\n
  4. What is nice-to-have vs must-have?<\/strong> If evaluation harness and failure taxonomy are non-negotiable for a technical cofounder, the market has put a price on reliability.<\/li>\n<\/ol>\n

    Working hypothesis while reading: companies buy the layer without which money or reputation is already leaking<\/strong> \u2014 not abstract \u201cmodel skill.\u201d<\/p>\n

    Five clusters that keep showing up<\/h2>\n\n\n\n\n\n\n\n\n\n
    Cluster<\/th>\nWhat they actually buy<\/th>\nTypical JD signal<\/th>\n<\/tr>\n<\/thead>\n
    Production agent ownership<\/td>\nHands-on EM \/ tech lead who deploys agents to customers<\/td>\nevals, guardrails, HITL, orchestration, \u201cnot just used AI tools\u201d<\/td>\n<\/tr>\n
    Applied AI leadership<\/td>\nDirector \/ Sr. Director who turns AI initiatives into delivery<\/td>\nmulti-team AI roadmap, architecture, GTM alignment<\/td>\n<\/tr>\n
    Evals as infrastructure<\/td>\nA dedicated prompts-and-evals function, not a side task for one IC<\/td>\neval suites, regression on model releases, mentoring product teams<\/td>\n<\/tr>\n
    Partner \/ field architecture<\/td>\nImplementation architecture through GSIs and cloud partners<\/td>\nreference architectures, partner-led pre-sales, enterprise readiness<\/td>\n<\/tr>\n
    Industrial \/ domain autonomy<\/td>\nC-level or Director with a mandate for autonomous \/ agentic shift<\/td>\n\u201cdrive transition toward agentic AI,\u201d OT\u2013IT, production environments<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n

    Each cluster below uses live wording. Where the employer is anonymized in the corpus, I keep it unnamed.<\/p>\n

    1. Production agent ownership<\/h3>\n

    A VC-backed AI startup (~$5M ARR) preparing for Series A is hiring three<\/strong> Software Engineering Managers for AI agents at once (EU\/UK, via Techmunity). The JD asks for a personal track record: agents in production; an explanation of how agents failed<\/em>; how reliability, evals, guardrails, and edge cases were handled. The squad owns orchestration, tool integrations, monitoring, human-in-the-loop (HITL) \u2014 plus strategic deployments that move customer automation from ~40% to 85\u201395%.<\/p>\n

    Read from the vacancy analysis: delivery used to sit with Customer Success and did not scale. The company is moving to engineering-led deployment before the round \u2014 eng must own customer outcomes, and the hiring structure has to show that to investors.<\/p>\n

    The same reliability emphasis shows up for a technical cofounder at an agentic startup (NL\/EU): must-haves explicitly include evaluation harness, regression prevention, error\/failure taxonomy, and monitoring \u2014 next to CI\/CD and safe releases. Without an eval loop, the candidate does not clear the bar.<\/p>\n

    2. Applied AI leadership<\/h3>\n

    Alongside startups, the market posts Sr. Director, Engineering \u2014 Applied AI<\/strong> roles (NAMER\/EMEA): AI engineering strategy across product lines, architecture, experiments \u2192 scalable outcomes, alignment with product\/design\/GTM. That is no longer \u201can ML team in the corner.\u201d It is the org layer that folds scattered AI features into managed delivery.<\/p>\n

    In enterprise pharma the same shift reads in Director, Enterprise AI Engineering<\/strong> (AstraZeneca, Gothenburg): five zones from Systems Intelligence to Product Incubation, a golden path for AI development, operational excellence for production ML, leadership over a PhD-heavy team. The role exists because the org is moving from a research-heavy setup to a production platform \u2014 and previously lacked a dedicated \u201cscience \u2192 delivery\u201d translator.<\/p>\n

    If you are an enterprise CTO still planning AI as \u201ctwo strong ML hires,\u201d JDs in this class already describe a different org chart.<\/p>\n

    3. Evals as infrastructure<\/h3>\n

    At a frontier lab (Anthropic), Prompt Engineer, Agent Prompts & Evals<\/strong> (SF; carded band on the order of $320\u2013405K) covers system and feature prompts, eval frameworks and automated suites, regression testing across product surfaces on model launches, and mentoring product teams. The vacancy read notes that prompting used to be fragmented across teams; an EM seat for the same line is open in parallel \u2014 the line is planned as an infrastructure function, not a lone IC.<\/p>\n

    That is the labor-market translation of the point I made separately on eval as a release criterion for AI features<\/a>: once model behavior is part of the product, demo-grade \u201clooked fine\u201d stops being acceptable \u2014 and headcount appears for the discipline.<\/p>\n

    4. Partner \/ field architecture<\/h3>\n

    The same lab hires Head of International Applied AI Architecture, Partnerships<\/strong> (London): a Partner Solutions Architect team, global system integrators (GSIs) and cloud partners, partner-led pre-sales, reference architectures, product feedback. Requirements include RAG, agentic systems, enterprise production readiness.<\/p>\n

    The stronger read is a platform pivot<\/strong>: marketplace, partner network, and field enablement outrun the lab\u2019s internal capacity to land every enterprise deployment with its own staff. Demand for applied architecture moves outward \u2014 to partners and to independent practitioners who can ship inside someone else\u2019s constraints. A neighboring layer is why AI skills need onboarding rather than a folder copy<\/a>: people and agent packages without a gate sprawl at scale.<\/p>\n

    5. Industrial \/ domain autonomy<\/h3>\n

    A quieter cluster is industrial AI at Series C. A CTO role (UK remote, \u00a3150\u2013190k + 1.5\u20132% equity) lists among responsibilities an explicit mandate to drive technical transition toward more agentic AI and autonomous optimization<\/strong>. That is a job duty, not a line from an investor deck. OT\u2013IT experience is often still nice-to-have: autonomy is already in the text, while OT\u2013IT scarcity has not yet forced the requirement into must-have.<\/p>\n

    For anyone who only watches consumer ChatGPT hype: Series C money is already writing agentic transition into C-level job text.<\/p>\n

    Limits of the method<\/h2>\n

    A cluster is not a stock forecast. A few constraints keep the read from becoming storytelling:<\/p>\n

      \n
    • Selection bias.<\/strong> The corpus is roles that entered the intel funnel (LinkedIn, recruiters, retained search). It is not a random sample of every AI vacancy on earth.<\/li>\n
    • One role without neighbors is a weak signal.<\/strong> A polished \u201cAI Engineer\u201d JD with no cluster can be a real hire \u2014 or roadmap decoration.<\/li>\n
    • Recruiter language lies.<\/strong> Strike \u201ctransform the industry.\u201d Keep measurable outcomes, ownership boundaries, and adjacent seats.<\/li>\n
    • Comp bands legitimize a discipline; they do not prove mass volume.<\/strong> A $320K+ seat on Agent Prompts & Evals says the lab treats the function as expensive infrastructure \u2014 not that thousands of such seats exist.<\/li>\n
    • Anonymous retained roles<\/strong> are useful as a pattern (3\u00d7 SEM) and dangerous as a \u201cnamed company story\u201d \u2014 some employers stay unnamed on purpose.<\/li>\n<\/ul>\n

      If after stripping marketing adjectives the JD has no failure mode, success criterion, or ownership boundary left, it is not yet a demand signal. It is a press release shaped like a vacancy.<\/p>\n

      What to do with this \u2014 CTO and candidate<\/h2>\n

      If you hire.<\/strong> Check your headcount plan against the clusters above. If the roadmap says \u201cagents in production\u201d and the hiring plan is only research ML plus prompt enthusiasts with no eval\/ops ownership, you are buying the demo the market already knows how to stage. Name an owner for the production agent path, the eval gate, and partner\/field architecture \u2014 even if that owner is fractional, not FTE.<\/p>\n

      If you are a candidate or independent architect.<\/strong> Watch where the price moved: reliability (evals, guardrails, failure taxonomy), applied delivery across teams, partner-facing architecture. That is closer to AI Architect \/ production systems than to \u201cI write good prompts.\u201d Positioning around systems that reach production \u2014 instead of dying in a pilot \u2014 matches what JDs already say, not what AI manifestos promise.<\/p>\n

      If you read lab news.<\/strong> Marketplace, partner network, enablement seats, and SOX\/compliance in one cluster are not random hires. They are the org chart of a company moving from model-as-a-service toward platform plus enterprise delivery. External demand for implementation architects rises in exactly those windows.<\/p>\n

      Bottom line<\/h2>\n

      In the parsed 2026 JDs, demand reads as buying the layers without which an agent or a model does not become a business: production ownership, eval infrastructure, Applied AI leadership, partner architecture, and sometimes industrial autonomy at C-level. \u201cMore people who know ChatGPT\u201d does not explain this corpus.<\/p>\n

      Hype sells a skill. Vacancies buy a control loop.<\/p>\n

      Assemble the cluster, strike the adjectives, keep outcomes and neighboring roles. What remains is the map of real demand.<\/p>\n","protected":false},"excerpt":{"rendered":"

      Hype says \u201chire AI.\u201d Job descriptions buy something else: production agents with evals and guardrails, Applied AI leadership, partner architecture, evaluation harness as a must-have. Method: read role clusters as an org chart under pressure.<\/p>\n","protected":false},"author":0,"featured_media":77,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[24,56],"tags":[26,62,60,61,29,59,57,58],"class_list":["post-78","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-ai-engineering","category-market","tag-ai-agents","tag-anthropic","tag-applied-ai","tag-demand","tag-enterprise","tag-evals","tag-hiring-signals","tag-vacancies"],"aioseo_notices":[],"aioseo_head":"\n\t\t\n\t\n\t\n\t\n\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t