WACE started from one problem: the factory has no operating system. From data standardization to AI autonomous control to equipment integration, we connect the whole path to the dark factory under one roof. We are always looking for people to walk that path with us.
Three years in, we are already profitable, and the product runs in real factories. At our size, you join as an early member — what you build shows up on the floor.
It would be easy to write that we want people who are good with AI, but that is not what we actually look at. What we look at is whether you catch it when AI is confidently wrong. In a field experiment run by Harvard Business School and BCG with 758 consultants, giving AI to people working on a problem AI handles poorly dropped their accuracy from 84.5% to 60~70%. Every participant was an expert in that field. Experience alone did not stop it. So on top of experience, we decided what else to look for — and what to look at first.
People who verify an AI answer with something other than the AI. Asking it again is not verification — challenged, a model tends to attach better-sounding reasons rather than back down. Analysis of real conversation logs found that the harder professionals fact-checked and pushed back, the harder the model pushed its case.
People who say how they measured it when they quote a result. We care less about "it got 30% faster" and more about "faster than what, measured how." Even experienced practitioners turn out to be badly wrong about their own productivity changes, so we trust the method more than the number.
People who have taken a problem outside their own training or track record and carried it — with AI — all the way to a real result. Every plant we walk into is one we have not seen before, running equipment we have not seen before. Starting without knowing and still getting to something that runs is exactly the experience we need.
People who decide what to hand to AI and what to keep in their own hands. Handing over everything and handing over nothing are equally risky. In the same experiment, the people who performed well worked in visibly different ways, but they all split the work into what they delegated and what they kept.
People who have built something that someone other than themselves actually uses. A working demo and a thing the floor opens every day are entirely different achievements. When enterprise AI fails to turn into results, the reason is usually not a weak model but that it never made it into real workflows.
Having written code yourself and made something actually run still matters — when it breaks on the floor, you are the one fixing it. But we place it third. If the two above are missing, no amount of development skill solves the problem we are facing. For the same reason we do not make "good with AI" a gate you have to clear — in the same experiment, on problems AI handles well, bottom performers gained 43% and top performers 17%. Make the fastest-acquired skill your bar, and you lose the people whose skills are not.
Measured it, crossed into it, got someone else using it, and above all checked it against something else — people who have all four are the highest bar we hold. And that bar does not include years of experience, the name of your last employer, your title, or your degree. All 758 people who slipped in that experiment worked at one of the top consulting firms in the world.
At WACE we favor engineers who go into the field and own the outcome end to end (FDEs) over developers who only write code. Even without a specific opening, if you are strong in one of the areas below, we would like to talk. Roles and terms are shared individually once you apply.
You go into the factory itself — diagnose the data, adapt and integrate our base system to that site, and validate it until it actually runs. Not just writing code, but owning the problem from definition to production, and turning what you learn on the floor back into the product.
Build the core of the product — the VEXPLOR Suite, Logic Studio and more. Large-scale manufacturing data, real-time control screens, and a platform on top of standard tables and an ontology — the foundation FDEs reuse in the field.
Build manufacturing AI and agents that sense, predict, and judge without hallucination. Work that runs from field data pipelines to model operations (MLOps) and simulation-based validation.
From business development and project management with manufacturing customers, to the content and marketing that get the product known, to the operations that hold a growing organization together.
| HQ (R&D) | Dongtan, Hwaseong, Gyeonggi — 150 Dongtan-yeongcheon-ro, Bldg. A, #1332–1335. Our product-development base, with a certified corporate R&D center. |
| Jeonju office | 737 Dongbu-daero 2F #201-J25, Deokjin-gu, Jeonju, Jeonbuk State |
| Team | 19 employees — 5 with a master's or PhD, built around engineers with 16 years on the manufacturing floor. |
| Employment | Full-time (probation included) · terms discussed individually by track and experience |
| Pay & benefits | Shared individually during the application stage, matched to track and experience. |
Leave your details on the inquiry form, or send your resume and portfolio by email. We review and get back to you individually. Even if there is no perfect opening right now, we keep you in our talent pool and reach out first when a role opens.
No perfect opening needed. Tell us what you have worked on and what you want to build.