Modelling techniques

Deep learning interview questions

Deep learning questions in data science interviews focus on the ideas that explain why modern networks train at all: backpropagation, why ReLU replaced sigmoid, what causes vanishing and exploding gradients, and what dropout and batch normalisation are doing.

More and more, interviewers also ask about attention and transformers: what queries, keys and values are, and why attention handles long-range context better than an RNN. You won’t be asked to derive everything, but you will be asked why, and a memorised answer tends to fall apart at the first follow-up.

Try one now

Friendly

Why did ReLU replace sigmoid as the default activation in hidden layers?

Free, no sign-up. The interviewer asks follow-ups on your reasoning, then rates your answer.

What the 14 questions cover

  • Neural networks and backpropagation8
  • Activation functions5
  • Vanishing and exploding gradients5
  • Transformers4
  • Attention: query, key, value3
  • Dropout and batch normalisation2
  • CNNs2
  • RNNs and LSTMs2
  • Transfer learning1
  • Encoder vs decoder1

A question can cover more than one topic, so the counts add up to more than 14.

How the practice works

  1. The AI interviewer asks the question and follows up on your answer, the way an interviewer would.
  2. Stuck? Ask for a hint. Each one gives a little more away, so you can get unstuck without being handed the answer.
  3. Get a written debrief. Four ratings, each from Needs work to Interview-ready, tied to what you actually said, and what to study next.

More modelling techniques questions

See all categories