AutoCAD interview questions

What a strong answer covers

• Model space is where the object is drawn at full size in real-world units. Nothing is scaled down while modelling, so measurements taken from the geometry are always true.
• Paper space, in a layout, represents the physical sheet at its real size. It holds the title block, the border and the annotation that belongs to the sheet.
• A viewport is a window in the layout looking into model space at a set scale. Different viewports on the same sheet can show the same model at different scales.
• Once a viewport scale is right it should be locked, so that zooming inside the viewport does not silently change the scale of the plotted view.
• Annotation size is decided by what must appear on paper, which is why annotative text and dimension styles are used: they present at the correct model size for each viewport scale automatically.
• A strong answer mentions that page setups tie the layout to a specific paper size, plot device and plot style table, so plotting the sheet is repeatable rather than reconfigured each time.

Where people lose the point

• ×Scaling the geometry itself to fit a sheet, which destroys the drawing as a source of measurements.
• ×Leaving viewports unlocked, so a zoom changes the plotted scale without anyone noticing.
• ×Placing sheet annotation and title blocks in model space, which makes multiple scales on one sheet unmanageable.

What a strong answer covers

• Layers separate a drawing into controllable categories so that visibility, colour, linetype, lineweight and plotting behaviour can be set for a whole class of objects at once.
• They are also how a drawing communicates to other people: a recipient who knows the layer standard can turn a discipline on or off, or plot a filtered version of the sheet, without editing the geometry.
• Layer 0 is special because geometry drawn on it inside a block inherits the layer properties of wherever the block is inserted. That is a useful behaviour for symbols intended to take on their host layer, and a source of confusion when it happens by accident.
• Properties should generally be set ByLayer rather than assigned to individual objects, so that a change to the layer changes everything on it. Object-level overrides defeat that and are hard to find later.
• Non-printing layers are how construction lines, notes to the drafter and setting-out geometry are kept in the file without appearing on the plotted sheet.
• A strong answer says that layer names should follow the office or project standard, notes that published standards for layer naming exist and differ between industries and countries, and asks which one applies rather than asserting one.

Where people lose the point

• ×Drawing everything on one layer, or on layer 0, which makes visibility and plotting control impossible.
• ×Setting colour and linetype per object instead of ByLayer.
• ×Inventing personal layer names on a project that has a defined standard.

What a strong answer covers

• A block groups geometry into a single named definition that can be inserted many times. Because the insertions reference one definition, editing the definition updates every instance, which copied geometry cannot do.
• Blocks also reduce file size relative to copying, and they make counting and scheduling possible because instances are countable objects rather than loose lines.
• A dynamic block adds parameters and actions so one definition can cover a family of variants: a door that can flip and change width, or a bolt available in several lengths, without a separate block for every case.
• Attributes attach text data to an instance: a tag, a mark number, a size or a manufacturer reference. That data can be edited per instance and extracted into a schedule or a spreadsheet, which is where much of the value is.
• Blocks intended to take on the properties of the layer they are inserted onto are drawn on layer 0 with properties set ByBlock, which is a deliberate technique rather than an accident.
• A strong answer mentions that a shared block library, kept under control rather than copied between projects, is what makes symbols consistent across an office.

Where people lose the point

• ×Copying geometry repeatedly instead of using a block, then having to edit every copy when the detail changes.
• ×Exploding blocks to make a small local edit, which destroys the link to the definition and the ability to count or schedule.
• ×Building a separate block for every size when a dynamic block with parameters would cover the family.

What a strong answer covers

• An external reference links a separate drawing file into the current one. The referenced geometry is not stored in your file, so when the source drawing changes, your drawing reflects the change when the reference reloads.
• A block copies its definition into the drawing, so it travels with the file but does not update when someone else changes their copy of the source. That distinction is the reason shared backgrounds are referenced and repeated symbols are blocked.
• Attach and overlay differ in nesting. An attached reference carries through when your drawing is itself referenced by another, while an overlay does not, which is how circular and duplicated references are avoided in multi-discipline coordination.
• Path type matters in practice: relative paths survive a project folder being moved or sent to another office, while a full path points at a directory that may exist only on one machine.
• Binding converts the reference into local geometry. It is appropriate for archiving, or for issuing to a recipient who will not receive the referenced files, and inappropriate during live coordination because it freezes the background and disconnects future updates.
• A strong answer mentions packaging a drawing set with its references and support files for transmittal, so that the recipient does not open a drawing with unresolved references.

Where people lose the point

• ×Binding or exploding references routinely to avoid path problems, and then working from a stale background.
• ×Using full paths, so references break the moment the project folder moves.
• ×Sending a drawing without its referenced files or fonts and assuming it will open as intended.

What a strong answer covers

• Styles are defined once and applied, not adjusted object by object. A drawing where every dimension carries its own overrides cannot be updated consistently and will not match the rest of the set.
• The template file is where styles belong. A properly built template carries the layer set, text and dimension styles, page setups, plot style assignment and title block, so a new drawing starts compliant instead of being corrected later.
• Annotative styles solve the scaling problem: the style stores the printed size, and the object displays at the correct model size for each annotation scale it is assigned to.
• Dimension style settings that get noticed on review include arrowhead type and size, precision and rounding, units and suppression of trailing zeros, and how the extension lines relate to the geometry.
• Consistency across the set matters more than any individual preference, because a set where three sheets dimension differently looks unchecked even when every value is correct.
• A strong answer mentions checking dimension values against the model rather than typing overrides, since a manually overridden dimension is a lie that survives every later revision.

Where people lose the point

• ×Typing a dimension value manually to make it read what it should, instead of correcting the geometry.
• ×Applying per-object style overrides that make a set-wide update impossible.
• ×Starting drawings from a copy of a previous job rather than a maintained template, so standards drift.

What a strong answer covers

• Plotted appearance is controlled by plot style tables. A colour-dependent table maps each colour to a plotted lineweight and colour, which is the traditional approach where object colour encodes pen weight; a named style table assigns styles to layers or objects directly, decoupling appearance from colour.
• A drawing uses one type or the other, and mixing conventions across a set is a common cause of sheets that print with inconsistent line weights.
• Page setups store paper size, plot device, plot area, scale, orientation and plot style. Saving them in the template, and applying the same named setup to every layout, is what makes a set print consistently.
• Publishing the whole set at once, rather than plotting sheets individually, avoids the situation where one sheet was plotted with different settings from the rest.
• Always check the output rather than assuming: a quick pass through the produced PDF catches missing xrefs, wrong scale, clipped title blocks and layers that should not have plotted.
• A strong answer mentions that what is issued should be a fixed-format file such as PDF, so the recipient sees what was checked rather than whatever their own settings would render.

Where people lose the point

• ×Plotting sheets one at a time with slightly different settings and issuing them as a set.
• ×Mixing colour-dependent and named plot style conventions within one project.
• ×Issuing without opening the resulting PDF to check it.

What a strong answer covers

• Nothing is edited silently. The issued drawing is a record of what was sent, so the change is made as a new revision with the previous issue retained and marked superseded.
• The change is identified on the sheet: a revision cloud around the affected area with a revision mark, and a line in the revision table giving the revision identifier, the date, a description of the change and who made and checked it.
• The revision is reissued through the same transmittal route as the original, so there is a record of who received which revision and when. Site working from a superseded sheet is the failure this whole process exists to prevent.
• Superseded files should be moved out of the live working area or clearly marked, so the current revision is unambiguous to anyone opening the folder.
• Related sheets have to be checked: a dimension change often affects details, schedules and other disciplines' drawings that referenced the same background.
• A strong answer states that revision numbering conventions, cloud and mark styles, and whether preliminary issues use letters and construction issues use numbers, are set by the office or project standard, and asks which applies rather than asserting one.

Where people lose the point

• ×Editing the issued file in place with no revision record, so nobody can tell which version site is holding.
• ×Clouding the change but not updating the revision table, or the reverse.
• ×Reissuing the changed sheet without checking the details and schedules that depend on it.

What a strong answer covers

• Work on a copy and keep the original untouched, because the received file is the record of what you were given.
• Run a file audit to find and fix structural errors, then purge unused layers, blocks, styles and other definitions, remembering that purge may need to be run more than once as nested items become unused.
• Check units before anything else geometric. A drawing created in different units, or with the insertion units unset, will scale wrongly when referenced or inserted, and that is far easier to find at the start than after a day of work.
• Remove duplicate and overlapping geometry, which is a common cause of both file bloat and incorrect quantity take-offs, and check for objects on frozen or off layers that will reappear later.
• Look for proxy objects and custom objects from other applications, which display but cannot be edited, and for missing fonts and shape files that will silently substitute.
• Rebuild rather than patch where it is faster to do so: if the layer structure is unusable, moving the geometry onto a compliant layer set from your template is often quicker than repairing it in place.
• A strong answer ends with a note back to the sender describing what was found, because a received drawing that does not meet the agreed standard is a project issue, not just a personal inconvenience.

Where people lose the point

• ×Working directly on the received file with no untouched copy retained.
• ×Purging aggressively without checking whether definitions are used by references or by the title block.
• ×Ignoring the units and insertion scale, then discovering the problem after everything has been referenced in.

What a strong answer covers

• Start with a maintained template that carries the layer set, text and dimension styles, page setups, plot style table, title block and standard notes, stored centrally rather than copied between machines.
• Keep a shared block and detail library under change control, so a corrected symbol propagates rather than surviving as five variants.
• Define the folder structure, file naming and revision convention in writing, since drawing names and revision letters are how everyone else finds and cites your work.
• Use the standards checking tools to compare drawings against a defined standards file, which catches layer and style drift automatically instead of by review.
• Agree the issue process: what is checked, by whom, what format goes out, and how transmittals are recorded.
• A strong answer acknowledges that industry layer naming and drawing numbering standards exist and vary by sector and country, and that the choice of which to adopt is a project or client requirement rather than a technical preference.

Where people lose the point

• ×Relying on individual habits and review comments instead of a template and a checkable standard.
• ×Letting each person keep a personal copy of the template, so versions diverge.
• ×Imposing a standard without checking the client's or the project's own requirements first.