GROUND TRUTH · MECHANICAL ENGINEERING
Mechanical Design Interviews Ask About Drawings, Not CAD Commands
The model shows what a part looks like. The drawing decides whether it can be made, checked and trusted.
An interviewer rarely asks how you made a fillet. They hand you a drawing and ask what it is telling you.
A student preparing for a mechanical design interview usually expects to be asked about the software: how to create a fillet, an extrude, a revolve, a pattern; how to build an assembly; how to run a quick simulation. That preparation is not wasted, but it is aimed at the wrong target. Most interviewers already assume a fresher can operate CAD software reasonably well, because the tool is not hard to learn and colleges spend real time on it. What the interviewer usually wants to know is something the software cannot show them directly: can this person read and produce a drawing that another human being, a machinist, an inspector, a supplier, can act on without guessing. That is why an interview so often moves away from the 3D model and toward a 2D drawing with numbers, symbols and tolerances on it.
The distinction is easy to miss because a CAD model and a drawing look like two views of the same thing. They are not doing the same job. The model describes geometry: the shape a part would have if it were made perfectly, to the exact numbers you typed in. The drawing describes intent: what has to be correct for the part to work, what is allowed to vary, and how someone downstream should verify it. A model has no opinion about which surfaces matter more than others. A drawing does, because a human decided which dimensions and tolerances control the part's function and which ones are just there for completeness.
What candidates prepare for vs what gets asked
| What people think | What it actually is |
|---|---|
| Knowing the CAD commands means I can design a part | The commands build geometry. Deciding what that geometry needs to guarantee is a separate skill, and it is the one interviewers usually test. |
| A finished-looking 3D model is proof of design ability | A model can look complete and still carry no manufacturing or inspection intent, no datums, no controlled tolerances, no surface callouts. |
| Tolerances are a formality added at the end | A tolerance is a design decision about cost, fit and function. Get it wrong and the part is either too expensive to make or does not assemble correctly. |
A drawing is where the design commits to something
Inside CAD, a dimension can be changed in seconds and the model updates instantly, so nothing about working in the software forces a candidate to think carefully about a number before typing it. A drawing removes that comfort. Once a dimension is toleranced and released, it becomes a commitment: a machinist will cut to it, an inspector will measure against it, and a supplier will quote a price based on how tight it is. A hole specified as 10.00 mm with no tolerance is not manufacturable, because nothing can be cut to an infinitely precise number; a hole specified as 10.00 ± 0.05 mm is a real instruction someone can follow and someone else can check. The interview question "what tolerance would you put on this dimension, and why" is not testing whether you remember a rule. It is testing whether you understand that every tolerance is a trade-off between how well the part needs to fit and how much it will cost to make that way.
GD&T exists because words are not precise enough
Geometric Dimensioning and Tolerancing looks, to a fresher, like a set of unfamiliar symbols layered on top of an already-finished drawing. In practice it is answering a question that plain dimensions cannot answer on their own: not just how big is this feature, but where exactly does it sit relative to everything else, and how much is it allowed to vary from perfectly straight, flat, round or positioned. Two holes can each individually be within their size tolerance and still not line up with the mating part if nothing controls their position relative to a shared reference. That reference is a datum, and choosing the right datums is itself a design decision: pick a surface that is not used to locate the part in assembly or in the fixture, and the tolerances built on it stop meaning what they are supposed to mean. An interviewer who asks "what datum would you use here, and why" is checking whether a candidate has thought about how the part will be held, made and measured, not just how it looks on screen.
The drawing is written for someone who never sees the 3D model
A machinist on a shop floor, an inspector running a CMM, or a supplier quoting a casting frequently work from the 2D drawing alone. They are not opening the CAD file and rotating the part to understand what you meant. If a critical dimension is missing, ambiguous, or contradicts another dimension elsewhere on the sheet, the part gets made wrong, not because the machinist made an error, but because the drawing failed to communicate. This is why interviewers sometimes hand a candidate an intentionally imperfect drawing and ask what is wrong with it: a missing tolerance, a dimension that conflicts with a note, a section view that does not show the feature it claims to, a surface-finish symbol on a surface that is not toleranced at all. Spotting these is not pedantry. It is exactly the skill that prevents an expensive part from being scrapped after it is already cut.
What the model shows vs what the drawing has to decide
| 3D CAD model | 2D manufacturing drawing |
|---|---|
| Shows the nominal shape at exact typed values | Shows what range of values is acceptable, and why |
| Every surface is equally "correct" | Some surfaces are functionally critical; others are not: the drawing says which |
| No statement about how the part is held or measured | Datums define how the part is located and inspected |
| Changes instantly, with no manufacturing consequence | Every callout has a cost and manufacturability consequence once released |
What a stronger interview answer sounds like
A candidate who has only practiced CAD commands tends to answer drawing questions by describing the geometry again: "it's a bracket with four holes and a slot." A candidate who understands drawings answers the question underneath the question: which of those holes has to align with the mating part, what happens if it is off by half a millimeter, and does that call for a tight position tolerance, a looser one, or a slot instead of a round hole to allow some play. That kind of answer signals that the candidate has thought about function, not just appearance, and it is a difference an interviewer can hear within the first thirty seconds of a response, long before checking whether the candidate can technically operate the software.
Surface finish and material notes carry the same weight as dimensions
Tolerances get most of the attention in a drawing conversation, but a complete drawing also specifies surface finish, material and heat-treatment callouts, and these carry just as much engineering weight even though they rarely come up in a first CAD tutorial. A surface that mates with a bearing or slides against another part needs a finish tight enough to control friction and wear, specified as a roughness value most freshers have never had to think about assigning themselves. A part that needs to survive fatigue loading may need a specific heat treatment noted directly on the drawing, because the same geometry in the same material behaves very differently depending on whether it was hardened, and skipping that note is not a minor omission. It can mean the part that gets manufactured is not the part that was analyzed. An interviewer who asks "what surface finish would you specify here, and why" is checking the same underlying skill as a tolerance question: whether the candidate understands what physically has to be true about the part, not just what shape it needs to be.
Revision history is part of the drawing, not an afterthought
A released drawing is rarely the final version forever: designs change, and every change to a drawing already in production needs to be tracked, dated and explained through a formal revision block, because a machinist or inspector working from an outdated revision can produce a part that is wrong in a way nobody notices until it fails to assemble or fails inspection. A candidate who has only worked on personal or academic projects usually has never had to think about this, because there was only ever one version of the drawing that mattered. In a real engineering organization, knowing which revision is current, understanding why a change was made, and being able to trace a dimension back through its revision history is a routine and unglamorous part of the job, and it is one more reason interviewers sometimes hand a candidate two versions of the same drawing and ask what changed and why it might matter.
Before your next design interview
- Pick one part you have designed and produce a real, fully toleranced drawing for it, not just a model.
- For every dimension you tolerance, be able to say what would go wrong if that tolerance were doubled.
- Identify the datums you would use, and explain why those surfaces (not others) locate the part.
- Find one deliberately flawed drawing example online or in a textbook and practice listing what is wrong with it.
- Practice explaining a tolerance decision in terms of cost and function, not just the rule that produced the number.
References and further reading
Background reading on the topic of this article. HireSetu is not affiliated with these organisations.