Author: MTF Institute Research Team
Independent review: MTF Institute Research QA
Evidence date: 11 October 2026 · United States
Technical report: MTF-CF-RR-2026-10-11-IIOT-PM
Executive summary
Industrial data becomes useful when someone can connect a signal to the right asset, determine what it means for production and maintenance, and make a decision that others can review. That work is distributed across several occupations. A reliability engineer may prioritize recurring failures; a condition-monitoring specialist may diagnose a machine-health alert; an industrial data engineer may make equipment information usable across factory systems; and a manufacturing engineer may turn automation performance into a validated operating change.
To examine how employers describe that work, we studied 107 current U.S. employer postings from 42 employers, retrieved on 11 October 2026. The postings were grouped by their primary duties into 53 reliability engineering, 33 industrial Internet of Things and operational technology (IIoT/OT), 17 smart factory operations, and four condition monitoring roles. These are analytical groupings for a structured, purposive sample, not standard job titles or a representative measure of U.S. hiring. A posting can touch more than one area, but it is assigned to only one primary family in these totals.
The evidence points to distinct kinds of outputs rather than one universal job description. Selected employers ask for asset and maintenance strategies, analyses of equipment failures, documented recommendations, connected equipment data models, automation requirements, dashboards and operational handoffs. They also specify different levels of authority: recommending a maintenance action is different from approving a production shutdown or changing live controls. The source pages often mix performed duties with required experience and preferred tools. This report keeps those categories separate and treats details that a posting does not state as unknown.
Employer concentration limits any numerical generalization. Divergent contributes 14 selected postings, SpaceX 12 and Anduril Industries nine. The sample contains genuine cross-employer variety, but several openings from one company can repeat the same terminology or recruitment template. A sensitivity check that caps each employer's total analytical weight at five gives a weighted denominator of 85. That check helps reveal concentration; it does not create new vacancies or convert the sample into a national survey. The findings below describe the selected postings and give direct employer examples. They do not estimate how many U.S. firms use a particular method, the number of future openings or the effectiveness of a technology.
Question and design
The research question was: What duties, work products, methods, systems, qualifications and decision boundaries do selected current U.S. employer postings explicitly describe for factory asset reliability and connected operations? The unit of analysis was one employer posting for a U.S. work location. Search results and third-party copies were discovery leads; the study required a direct employer page showing an identifiable role, location, substantive duties and evidence that applications were open at retrieval. Public postings can be edited or removed later, so the collection date matters.
The four families cover different parts of an occupational field. Reliability engineering involves equipment performance, maintenance strategy and failure elimination. Condition monitoring involves diagnosis and communication based on asset-health evidence. IIoT/OT work involves industrial connectivity and information that supports factory operations. Smart factory operations involves connected production systems, operating performance and maintenance or engineering handoffs. A job qualified for the study through its actual responsibilities, not because its title contained words such as “digital,” “automation” or “reliability.” Generic cloud software reliability, consumer IoT, sales, unrelated production planning, jobs outside the United States and roles centered only on physical repair or field measurements fell outside the core professional sample.
Where an employer supplied a requisition identifier, that identifier and the employer were used to distinguish postings. For pages without one, the employer URL, title and location supplied the identity check. Distinct URLs may represent distinct employer pages without proving distinct employer-issued requisition numbers. A multi-location page was treated as one posting unless the employer published separate openings. The study retained derived codes and short supporting expressions rather than full job-page copies. Each posting was examined for duties, named outputs, methods, observable work with other people, tools or systems, qualifications, work rhythm, interfaces, decision rights and escalation. When wording did not state a fact, the value remained unknown rather than being counted as a negative answer.
This design is intentionally more demanding than counting search hits. It is also narrower than a national labor-market survey. Employers choose what to advertise and how much detail to include. Their advertisements can describe a desired profile more fully than the eventual job. The selected pages span different industries and levels of seniority, but selection was purposive and employer activity was uneven. The sample can support careful statements about what particular employers explicitly requested. It cannot establish U.S. prevalence, hiring growth, pay outcomes or the performance of any equipment or software system.
The way a requirement is written changes what can be concluded from it. If a posting says a person will investigate recurring equipment failures, it describes assigned work. If it calls for a documented analysis or maintenance plan, it names an output. If it asks applicants to have used a diagnostic method before, it describes candidate experience. A technology listed among optional advantages is weaker still as evidence about the eventual task. This distinction prevents an attractive list of brand names or technical phrases from being mistaken for an account of what a worker does on a normal shift. It also helps preserve legitimate alternative entry routes: some employers accept a degree or equivalent experience, while others specify a credential or prior industrial exposure for a particular role.
Silence is another important feature of vacancy evidence. A role that says nothing about sensor quality does not necessarily ignore it; the employer may leave operational details to an interview or internal procedure. A page that omits an escalation route does not imply that an engineer can approve a shutdown alone. For that reason, this report uses explicit examples to explain each kind of work and avoids turning unstated fields into claims of absence. The most reliable comparison is often between named outputs and permissions, with the missing parts left open rather than supplied by intuition.
Four related kinds of work
Reliability engineering
The 53 reliability engineering postings form the largest primary family in this sample. Their common analytical concern is the condition and performance of production assets, although their scopes vary from plant-level strategy to individual equipment systems. The Cargill Reliability Engineer opening in Blair, Nebraska describes work on recurring equipment defects, predictive-maintenance activity and guidance on critical spares. The Sandvik Coromant Reliability Engineer opening connects asset criticality, maintenance plans and failure analysis. These examples show that reliability work can produce decisions about where maintenance effort should go, not merely collect a condition reading.
Other postings frame reliability as a management system. AstraZeneca's Philadelphia Reliability Engineer opening describes assessments and maintenance-program improvements as well as monthly performance reviews. West Fraser's mill Reliability Engineer opening ties machine-failure investigations and condition routes to downtime trends. At the Antares equipment reliability opening, the output includes an equipment roadmap, a maintenance-system hierarchy and handoff material. These are different production environments. Their named methods should be attributed to the particular page, not read as a mandatory package for every reliability role.
IIoT and OT engineering
The 33 IIoT/OT postings connect factory equipment and operational information. In the General Dynamics Mission Systems manufacturing systems opening, the work includes equipment context and integration among machine, supervisory and enterprise systems. The Mach Industries factory-software opening describes equipment data models, industrial integration patterns and dashboards. The RTX Raytheon digital factory opening similarly identifies factory connectivity and historian-based information as part of its scope.
These postings illustrate a practical problem: an equipment signal is hard to act upon if its source, operating state or production context is unclear. An integration engineer may therefore produce a model, interface or validated data flow that others can use. That does not make every OT engineer a predictive-maintenance analyst. The Redwood Materials staff controls opening, for example, concerns factory controls and safety-critical software. It is relevant to connected industrial systems, yet its responsibilities should not be redescribed as read-only machine-health diagnosis. OT architecture, controls engineering and reliability analytics overlap at interfaces while retaining different duties and permissions.
Smart factory operations
The 17 smart factory operations postings focus on production-system performance and the introduction or improvement of connected manufacturing capabilities. The Seagate senior factory automation opening describes system requirements, acceptance criteria, reliability work and performance measures for automation in a fabrication setting. The Anduril production manufacturing engineer opening connects new manufacturing systems and equipment with validation, documentation and work instructions. These are outputs that let a production organization decide whether an automated change works as intended.
This family shows why “smart factory” should be analyzed through operational work rather than used as a technology slogan. A dashboard can make losses visible, but a role may also have to confirm what the metric represents, agree on an acceptance test, document the change and coordinate with operators or maintenance. A factory software engineer may own a data interface; a manufacturing engineer may own a production process or equipment handoff. The job pages determine which responsibility is actually assigned.
Condition monitoring
The four condition-monitoring postings are a small and concentrated family. The I-care USA Predictive Maintenance Analyst opening describes analytical interpretation of wireless machine-condition streams and reports to clients. The I-care Predictive Maintenance Engineer opening similarly describes diagnosis and recommendations. Its condition-monitoring program manager opening operates at a different level, coordinating a client program and its decisions. Because several selected roles are from the same provider, their similar wording should not be mistaken for broad, independent employer confirmation.
The distinction between interpreting a condition stream and physically collecting a measurement is important. Both can be valuable occupations. This study's condition family focuses on analytical and program decisions; field collection and physical intervention are adjacent work. The small subgroup is enough to show examples of a specialty, not enough to characterize the whole U.S. condition-monitoring labor market.
Duties, outputs, methods and tools
A useful reading of a vacancy starts with its verbs and named work products. “Develop a maintenance plan” is a duty and output. “Experience with vibration analysis” may be a qualification. “Familiarity with a platform” may be preferred experience. Those statements answer different questions and should not be combined into one tally of what employees do. The Advanced Technology Services Maintenance Reliability Engineer posting illustrates the issue. It describes reliability planning and failure work among principal responsibilities, while listing several diagnostic technologies under desirable knowledge and skills. The technologies are part of the employer's stated candidate profile; their appearance there alone does not prove that each is a recurring task.
In reliability roles, a work product can be a criticality ranking, maintenance strategy, investigation finding, corrective-action proposal or revised plan. Cargill and Sandvik Coromant provide direct examples of asset and maintenance decisions. A technician may act on those decisions, while the engineer may be responsible for analysis, priorities, documentation or follow-through. A posting that merely names a method in a preferred section cannot establish this chain of work.
In condition roles, the analytical chain is a signal, an asset attribution, a diagnosis, a recommendation and communication to a responsible party. The I-care analyst opening explicitly centers interpretation of condition streams. Whether a diagnosis leads to a repair depends on plant constraints and decisions beyond the alert. Likewise, a model-generated score cannot by itself show that an asset is safe to operate or that intervention improved reliability. The posting establishes the requested work; it is not an evaluation of the employer's technology.
In data and factory operations roles, outputs include equipment identifiers and context, data flows, interface standards, operating views, system requirements and acceptance criteria. General Dynamics Mission Systems, Mach Industries and Seagate supply examples. Industrial protocols, historians, supervisory systems, manufacturing execution systems and maintenance systems can appear in these postings, but each platform has to be read in its own source context. For example, Seagate places SQL, Python and Power BI in a preferred section. Cargill lists SAP as preferred. Neither page proves that those products are universally required or should be endorsed over alternatives.
The behavioral dimension is similarly concrete. Employers may ask a person to explain a diagnosis to a client, coordinate a corrective decision with maintenance, document an automation change or train operating staff. These are observable tasks when tied to an artifact or handoff. Generic wording such as “good communication” is weaker evidence that a role owns a maintenance workflow. Antares names maintenance-system records and handoff material. I-care's analyst names client alerts and reports. Those concrete outputs are more informative than a general skill adjective.
Observed work themes in the selected postings
A theme was counted only when the employer assigned a performed duty or named work product with source-specific support. A technology named only in a required or preferred candidate profile was coded separately. Each posting could support more than one theme and counted at most once within a theme. Not evidenced means the advert did not support that code; it does not mean the employer never uses the practice.
| Assigned duty or named output | Raw mentions within 107 selected postings | Employer-cap-five weighted mentions (total sample weight 85) |
|---|---|---|
| Production-asset reliability strategy | 56 | 48.2 |
| Pre-failure condition-data diagnosis | 12 | 11.3 |
| Factory OT data integration | 28 | 21.9 |
| Production performance and failure analysis | 66 | 55.6 |
| Documented maintenance work product or specified handoff | 35 | 33.7 |
The cap-five calculation gives each posting from an employer with more than five selected openings a weight of 5 / that employer's selected posting count; every other posting has weight 1. The weights sum to 85 across the current sample. This sensitivity check reduces the influence of repeated openings from one employer. It is not a second vacancy sample or a probability-weighted estimate of the U.S. labor market.
Reliability strategy and performance or failure analysis were both evidenced in 50 selected postings, or 42.9 weighted mentions under the same cap. This co-occurrence describes the selected pages only. It does not establish that either task is typical of every U.S. factory or every role in this occupational field. The complete posting roster and coding flags accompany the archived report so that the denominator and source links remain inspectable.
Qualifications, timing and authority
Required and preferred qualifications vary across the selected occupations. Cargill's Blair posting gives an education or equivalent-experience route and lists some predictive technologies and SAP as preferred. The Danone Automation and Operations Technology Engineer posting describes an engineering-degree-and-experience route or longer equivalent experience. I-care's analyst opening describes analytical ability in a different experience profile. The contrast is a reminder to preserve the exact alternatives in each posting; a degree reference should not be converted into a universal entry barrier, and preferred knowledge should not be reported as required.
Work rhythm also depends on the role. Danone's OT engineer opening specifies daily OT health checks and monthly reporting. AstraZeneca's Philadelphia reliability opening names monthly maintenance reviews. Cargill states a regular weekday schedule with occasional weekend or on-call work. Where a posting does not state a cadence, it is not sound to assume continuous monitoring, a daily review or a standard reporting period.
Decision rights are no less specific. An engineer may identify high-risk equipment and recommend a work plan; a manager may set priorities; a separate authorized person may approve a production interruption. Danone names escalation of serious OT incidents to automation, cybersecurity and plant leaders. Cargill directs complex issues to appropriate staff. The I-care program manager posting includes stop-work authority in its own employment setting. None of these statements grants authority in another plant. If a vacancy does not describe its approval route, that route remains unstated.
Concentration, geography and interpretation
The 107 postings come from 42 employers, with several openings from the same employers and some similar job templates. Divergent's 14, SpaceX's 12 and Anduril Industries' nine selected openings together account for 35 postings. A company running a large recruitment campaign can therefore influence raw counts more than many single-opening employers. As a sensitivity check, each posting received a weight of one unless that employer had more than five selected postings; those employers' postings were weighted proportionally so that each employer contributed at most five. The resulting weighted denominator is 85. It is a way to inspect concentration within this sample, not a second headcount or a statistical correction that makes the sample nationally representative.
The study's principal geography is the United States, anchored in the selected employer pages' U.S. work locations. The examples include manufacturing settings in food, paper, electronics, aerospace and other sectors. This range gives different operational contexts, but it does not show how often a practice occurs across the national manufacturing base. A posting with a broad U.S. or multiple-site location should be interpreted at the location precision that its employer actually supplies. City-level claims are unnecessary when the page itself is ambiguous.
Broader U.S. occupational sources provide context without changing this denominator. The Bureau of Labor Statistics' industrial engineers profile addresses production-system improvement across a much wider occupation. Its industrial machinery mechanics and maintenance workers profile covers hands-on equipment maintenance that interfaces with, but is not identical to, the professional analytical roles examined here. NIST's advanced manufacturing services provide domain context for connected factory information. None of these sources supplies a count of the exact roles in this vacancy sample.
Employer postings are copyrighted source material. This report paraphrases them and links to the direct pages instead of reproducing their descriptions. Named standards, methods and vendors appear only to identify the work an employer described, without implying endorsement. Likewise, occupational references to physical repair, commissioning or control-system changes should not be interpreted as permission for an unqualified person to perform them. Physical intervention, energized work, live PLC or control changes and safety-system actions require the employer's qualified personnel, site procedures and change controls. Analytical findings and recommendations should travel through the responsible operating and maintenance teams.
MTF Institute's separate article, What Changed in U.S. Industrial IoT and Predictive Maintenance in 2026, examines dated industry and public-sector developments. It is a current-change study, with a different evidence base from these employer postings. A vendor announcement or public draft guidance is not a vacancy, and a vacancy does not establish broad adoption of a newly announced product or practice. The two forms of evidence can inform different questions without merging their counts or conclusions.
What the sample can and cannot say
The selected postings reveal a practical sequence across related jobs: produce trustworthy equipment information; connect it to an asset and operating context; investigate a performance or condition concern; document an appropriate recommendation; and deliver it to the person with authority to decide or act. Different roles own different parts of that sequence. Reliability engineers can own strategy and failure analysis. Condition specialists can own diagnosis and alerts. IIoT/OT engineers can own the data path and its integrity. Factory operations engineers can own acceptance criteria and production performance. This is a synthesis of the role boundaries illustrated by employer postings, not a claim that every job follows the entire sequence.
The study is limited by its purposive selection, the concentration of openings at some employers, uneven detail across pages and the temporary nature of public postings. A missing requirement is unknown rather than absent. A method listed as preferred experience should not be treated as a performed duty. The four-posting condition-monitoring subgroup supports examples, not a broad ranking of that specialty. The corpus does not measure applications, hires, wages, future openings, equipment outcomes or market-wide prevalence. Its value is a sourced account of the work that selected U.S. employers chose to describe at a specific date—and a clear view of where reliability, condition analysis, factory data and operating decisions connect or remain distinct.
Full source and coding appendices
The archived, searchable technical report PDF contains all 107 direct employer-page links, the source-by-source theme flags, methods and limitations. Its version DOI is 10.5281/zenodo.23300286. Vacancy pages may change after the 11 October 2026 retrieval date.
Continue learning
Apply the evidence from this report through MTF Institute's Professional Certificate in Industrial IoT, Predictive Maintenance & Smart Factory Operations. The programme turns the identified capabilities into structured theory, guided AI practice and reusable workplace artifacts.