Production engineering
Production research, evaluated on the line, not the pilot
How process judgment shows when a method has to survive a full shift, and the standard the founding cohort will hold it to. Everything below is a draft in public, on purpose.
What counts as evidence of production research skill
This discipline is defined by a gap that every practitioner knows and few papers describe: the distance between a result that holds in a pilot cell and one that holds across three shifts, two suppliers and a seasonal change in humidity. Evidence of skill is evidence about that gap — a capability study with its assumptions exposed, a changeover redesign measured before and after, a yield improvement that survived the quarter after the consultant left.
Process data is the discipline's native artifact and its native trap. Control charts, OEE breakdowns and scrap analyses are abundant; what distinguishes a researcher is whether they treat a signal as causal only after they have earned it. An engineer who ran the confirmation trial, or who says plainly that the improvement coincided with a tooling change they cannot separate out, is showing the skill the rubric is looking for.
Scale-up reasoning is evaluable even without proprietary numbers. Why a batch process resisted continuous conversion, which constraint actually governed throughput, what the line did when the bottleneck moved — these can be argued from ratios and mechanisms alone, and an evaluator can read the quality of the argument without ever seeing the plant.
The rarest visible skill is honesty about sustainment. Most reported improvements decay. A researcher who reports the twelve-month number as well as the launch number, and who explains what eroded, is doing something the published literature in this field almost never does.
The production research evaluation rubric, first draft
This rubric reads process work the way an operations review reads a claimed improvement: for whether the gain was real, whether it was caused, and whether it lasted.
- Measurement-system credibility
- Before any improvement is claimed, the measurement is shown to be trustworthy — gauge capability, sampling scheme and definition of the metric are stated, and drift in the measurement is distinguished from drift in the process.
- Causal attribution
- Improvements are separated from confounds — concurrent changes, learning effects, seasonality, selection of the comparison window. Where attribution could not be isolated, the author says so instead of implying a mechanism.
- Scale and transfer reasoning
- The argument for why a result holds beyond the cell it was found in is made explicitly, naming the constraint that governs at scale rather than assuming the pilot's constraint carries over.
- Sustainment evidence
- Performance is reported over a horizon long enough to decay, with the standard work, training or control that holds the gain identified — and any erosion reported rather than trimmed from the window.
Founding production evaluators will attack this draft first — including against their own improvement reports and capability studies.
What founding production evaluators will do
Take the rubric apart: where it demands data a plant will never release, where it mistakes methodology fluency for process judgment, where a dimension rewards the tidy write-up over the harder result.
Run calibration rounds on public work — published case studies, standards submissions, open manufacturing datasets, conference process papers — scoring independently and comparing spreads, so the first published scores come with known uncertainty rather than false precision.
Decide what the community does about survivorship: the field's published record is overwhelmingly successes, and a rubric that ignores that will reward the same bias it inherits.
Who this is for
The founding cohort is looking for engineers whose process judgment is already under load, wherever it is exercised:
- Manufacturing and process engineers whose strongest work is an internal improvement no journal will publish.
- Quality and reliability engineers who want capability reasoning judged rather than certification counted.
- Industrial-engineering and operations researchers who want their models read against what the line actually did.
- Supply-chain and continuous-improvement practitioners whose results are documented and measurable.
Who this is not for
Self-selection matters more than any filter we could write, so here is the honest version:
- Anyone after a credential for its own sake — the founding stage produces standards, not badges.
- Anyone whose case for expertise is a methodology certification rather than a result they can reason about.
- Anyone uncomfortable having their evaluation accuracy tracked — the calibration record is the point of the design.
- Anyone who needs a live scoring platform today; the mechanics on this page are in design, and the tense is deliberate.
Apply to evaluate production engineering
Production engineering is pre-selected on the application. Link to work we can read — a capability study, a process case study, a standards submission, an ORCID or repository profile.