What does a Software Engineer interview actually cover?
A Software Engineer interview on TheInterviews is the broadest of the five role families and the one most often used as a first screen: 15 of its 37 graded interviews ran as workforce screening and 13 as theory-only technical sessions. It draws the widest candidate pool we have — 30 distinct people across 37 interviews, so almost everybody who sits it sits it once — and it is graded primarily on technical knowledge, communication, problem solving and coding. It is also the role family with the toughest outcome mix, which is a fact about a first-screen population, not about software engineers.
Drawn from 37 graded Software Engineer interviews by 30 people, between 26 June 2026 and 28 August 2026 — part of the same 177 graded interviews the published rubric draws on. Rubric figures as of 28 August 2026; role breakdown as of 30 August 2026.
Which interview types were actually run
| Interview type | Interviews |
|---|---|
| Workforce screening | 15 |
| Technical theory only | 13 |
| Technical (theoretical) | 3 |
| Behavioural | 2 |
| Technical coding | 1 |
| Portfolio review | 1 |
| Finance analytical | 1 |
| Customer service | 1 |
Which dimensions the grader actually emitted
| Dimension | Interviews |
|---|---|
| Technical knowledge | 33 |
| Communication | 20 |
| Problem solving | 18 |
| Coding | 18 |
| Conceptual clarity | 15 |
| Trade-off analysis | 15 |
| Edge cases | 15 |
Dimensions appearing on fewer than two interviews are omitted. Every per-interview dimension is rolled up onto the five canonical dimensions described on the rubric page.
How these interviews came out
- YES
- 3
- of 37 graded interviews
- MAYBE
- 10
- of 37 graded interviews
- NO
- 24
- of 37 graded interviews
Read the mix as the grader's output, not as a verdict on the people. Our own evaluation, published at /methodology/scorer-compression, found the scorer separates behaviourally distinct candidates far less well than an independent rater does, and the published rubric at /scoring lists that and the other measured limits in full. The transcript is published beside every score; where the two disagree, the transcript is the one that is true.
This is the role people sit once, as a first screen
37 graded interviews from 30 distinct people is the widest ratio of the five families — near enough one interview per person. Every other role on this list is dominated by repeat practice; this one is dominated by first attempts. That single fact explains most of what is unusual about its numbers.
It also means the interview is doing screening work: 15 of the 37 ran as workforce screening, the highest count of any role here. A screen is broad and shallow by design — it establishes whether the fundamentals are there before anyone spends an hour on depth.
The outcome mix is the hardest of the five, and it is a sampling artefact
Of 37 graded interviews, 3 ended on YES, 10 on MAYBE and 24 on NO. That is the toughest mix of the five roles and we are publishing it rather than hiding it, but read it correctly: this is overwhelmingly a population of people taking an unfamiliar interview format for the first time, on a broad screening rubric, with no second attempt in the data. The roles further down this page with kinder mixes are the ones where the same handful of people practised repeatedly.
A first-attempt screening distribution is not a statement about software engineers, and it is not a statement about you. It is the strongest argument on this page for sitting a second one.
Communication is graded almost as often as technical knowledge
Technical knowledge appears on 33 of 37 graded interviews and communication on 20 — the highest communication share of any role in this set. In a screening format the interviewer has to establish breadth quickly, so an answer that arrives structured, scoped and with its assumptions stated is worth materially more than the same content delivered as a stream.
The cheapest available improvement for this interview is not more knowledge. It is saying what you are about to do before you do it, and saying what you would check if you had more time.
Software Engineer interviews — common questions
What is asked in a Software Engineer interview on TheInterviews?
Most often a broad screen: of 37 graded software engineer interviews between 26 June and 28 August 2026, 15 ran as workforce screening and 13 as theory-only technical sessions, with single instances of behavioural, coding, portfolio and other formats. It establishes breadth of fundamentals rather than depth in one area.
What is graded in a Software Engineer interview?
Across 37 graded interviews the grader most often emitted technical knowledge (33), communication (20), problem solving and coding (18 each), then conceptual clarity, trade-off analysis and edge cases (15 each). Communication carries more weight here than in any other role family in our data.
Why did so many Software Engineer interviews end on NO?
Of 37 graded interviews, 3 ended on YES, 10 on MAYBE and 24 on NO. This family is almost entirely first attempts — 30 distinct people across 37 interviews — sitting a broad screening rubric in an unfamiliar format. Role families where the same people practised repeatedly show much kinder mixes. The scorer also has documented limits, published in full with the rubric.