Skip to content
Sign inTelegram

An interview that shows what you missed

Answer by text or voice; the AI checks your answer against a rubric point by point and asks a follow-up. 5000+ questions from real interviews across 20 stacks.

Try it free Pricing

Free: 25 Abyss Token for AI interviews and 10 Quiz sessions. No card needed.

cpp / simulator / move semanticsC++Middle

How does std::move differ from moving, and what does it actually do?

Your answer

std::move moves nothing — it is a cast to an rvalue reference that lets overload resolution pick the move constructor or move assignment. The actual move is done by whatever function ends up being called.

Rubric breakdownGrade 6 / 10
  • A cast, not a move counted: static_cast<T&&>, the callee does the move
  • Who performs the move counted: move constructor or move assignment
  • State of the moved-from object partial: valid-but-unspecified mentioned in passing
  • When move hurts missed: return std::move(local) defeats NRVO
Follow-up

What is guaranteed about the object after the move — and when does std::move on a return get in the compiler’s way?

Three steps to a breakdown

A session is several questions in a row with a breakdown after each one, not a single test at the end.

  1. Pick a stack and topics

    20 stacks, from C++ to system design. Any topics from the pool or the whole set; junior, middle or senior.

  2. Answer like in a real interview

    By text or voice, in your own words. Unclear wording? Ask before you answer, like you would an interviewer.

  3. Get a point-by-point breakdown

    What counted, what was partial, what missed. A follow-up question instead of a flat “wrong”.

What people ask

How does the AI grade an answer?
Against a rubric. Every question has a list of points, and your answer is checked against each one: counted, partial, missed. The server computes the grade from those points; the model does not eyeball it.
Text or voice?
Both. A voiced answer is transcribed and graded against the same rubric as a typed one.
Which stacks?
Twenty: C++, Unreal Engine, Go, Python, C#, Java, JavaScript, TypeScript, QA, PHP, Kotlin, Swift, SQL, system design, data analytics, ML, security, DevOps, system analysis, management.
What is free?
After sign-up: 25 Abyss Token for AI interviews and 10 Quiz sessions. No card needed. 1 Abyss Token is roughly one AI-graded question; the rest comes with the subscription.
How is this different from asking ChatGPT?
ChatGPT does not know what your stack gets asked in interviews and does not check your answer against a rubric. Here the questions come from real interviews, grading follows fixed points, and weak topics come back in later sessions.

Weak topics stay lit until you close them

Every answer is recorded per stack: topic → concept → question. The weak spot stays highlighted until you answer it with confidence.

Questions and rubrics are written by Kirill Valitskii, a backend and gamedev engineer with 6+ years on Go and C++.

Try it free Pricing