How to screen resumes for engineering roles

The method in five steps

1. Write three must-have criteria and two nice-to-haves before reading anything.
2. Read every resume in the same order: most recent role, scope, then everything else.
3. Give each one the same few minutes.
4. Record a one-line reason for every reject.
5. Review your pass rates monthly against those reasons.

Write the criteria first

If criteria are not written down, they get invented while reading, which means they get invented differently for each candidate and in a way that tends to favour the familiar. Three must-haves is the right number — more and everything fails, fewer and nothing filters. They should be observable from a resume: “has shipped and operated a production service”, not “strong engineer”.

Read for scope, not keywords

• Scope: what did they own, how big was it, who else was involved? A sentence describing a decision beats a list of technologies.
• Evidence of shipping: things that reached users and survived contact with them.
• Progression: increasing responsibility over time, which matters more than the absolute level reached.
• Tenure pattern: look at the shape, not any single short stint. One short role explains itself; five is a pattern worth a question rather than a rejection.

Signals that feel informative and are not

• Employer brand. Tells you who passed another company's filter, which may have been testing something you do not care about.
• University. Among the strongest bias vectors available and weakly predictive for experienced hires.
• Keyword density. Filters for people who tailor resumes, a skill unrelated to engineering.
• Formatting and typos. Weak signal at best, and it penalises non-native speakers and people without time rather than careless engineers.
• Employment gaps. Carry essentially no information about ability, and screening on them disproportionately removes carers and people who have been ill.

Timebox and keep it even

Decide how long each resume gets — two or three minutes is typical — and keep it the same. The usual failure is spending ten minutes on an interesting resume and thirty seconds on an unfamiliar one, which converts curiosity into a selection mechanism. If you cannot decide within the timebox, that is a pass to the screen, not a reason to keep reading.

Record a reason for every rejection

One line is enough: which must-have was missing. This takes seconds and does three things — it forces the criteria to be applied rather than felt, it gives you data to review later, and it makes it possible to answer a candidate who asks. Reviewing a month of reasons is usually uncomfortable and always informative: most teams find one criterion doing almost all the rejecting, and it is rarely the one they thought mattered most.

Calibrate with a sample before you start

Before screening a new pipeline, take ten resumes and have two people screen them independently against the written criteria, then compare. Disagreement at this stage is cheap and extremely informative: it almost always exposes a criterion that reads clearly but is being applied in two different ways. Fixing the wording once, before the other two hundred applications, is worth considerably more than reviewing rejections afterwards.

Anonymise what you can, early

Removing names, photos, universities and dates helps at this stage specifically, because resume screening is where demographic proxies are most available and least checked. It stops helping once interviews begin, so treat it as a targeted fix rather than a philosophy. See how to reduce interview bias.

Use the screen to resolve doubt, not the resume

The most expensive screening error is rejecting on an ambiguity that a thirty-minute call would have settled. A resume that is hard to read, written in a second language, or vague about scope is not evidence of a weak engineer — it is evidence of a weak resume, which is a different skill. When the question is “I cannot tell from this”, the answer is a screen. Reserve rejections for cases where a must-have is clearly absent.

Screening is not assessment

A resume can tell you someone has probably done work of roughly the right shape. It cannot tell you whether they did it well, and trying to extract that is how good candidates get filtered on presentation. Pass anyone plausible to a thirty-minute screen, which is cheap and far more informative. See interviewing vs screening.

Where Interviewer Mode fits

Disclosure: we make CoPilot Interview, which includes Interviewer Mode for the hiring side and a real-time assistant used by candidates. We have an interest in you using the first, and our view of the second is that interviews should be designed to measure something an assistant cannot fake. Weigh the advice below accordingly.

Frequently asked questions

How do you screen resumes for engineering roles?

Write three must-have criteria before reading anything, read each resume in the same order and for the same few minutes, look for scope and evidence of shipping rather than keywords, and record a one-line reason for every rejection so you can review what you are actually selecting for.

What should I look for on an engineering resume?

Scope of ownership, evidence that things shipped and were operated, and progression in responsibility over time. A sentence describing a decision the person made is worth more than a list of technologies.

Should I screen out candidates with employment gaps?

No. Gaps carry essentially no information about engineering ability, and screening on them disproportionately removes carers and people who have been unwell. If the gap is relevant to the role, it is a question for the screen, not a rejection.

Does the university on a resume matter?

Rarely, and it is one of the strongest bias vectors in hiring. For experienced candidates it predicts very little. Unless the role genuinely requires a specific qualification, it is better removed from the screening view entirely.