Topic 13: Interview Q&A

What You'll Learn

This topic provides 100+ interview questions covering:

  • Classical ML
  • LLM training and inference
  • Parameters and optimization
  • Bias, variance, regularization
  • Theory and implementation

Categories

  1. Classical ML (20 questions)
  2. LLM Fundamentals (20 questions)
  3. LLM Inference (20 questions)
  4. Training Techniques (15 questions)
  5. Optimization (15 questions)
  6. Regularization (10 questions)

See INTERVIEW_QA.md for complete questions and answers.

Core Intuition

This topic is a practice tool, not just a reading list.

The real skill it trains is:

  • answer quickly
  • answer correctly
  • survive follow-up pressure

That means the best way to use it is active recall, not passive reading.

How to Use This Topic

Pass 1: Recognition

Read the question and make sure you understand what concept it is targeting.

Pass 2: Recall

Hide the answer and answer it out loud from memory.

Pass 3: Compression

Try giving:

  • a 30-second answer
  • a 2-minute answer

Pass 4: Follow-Up Pressure

Ask yourself:

  • what assumption did I make?
  • what edge case did I skip?
  • what would the interviewer ask next?

Common Failure Modes

  • over-explaining before stating the core answer
  • jumping to formulas without intuition
  • answering the main question but failing the first follow-up
  • memorizing words without understanding the mechanism

Edge Cases and Follow-Up Questions

  1. Can I explain this without jargon?
  2. Can I compare it to the closest alternative?
  3. Can I state the assumption clearly?
  4. Can I adapt the answer when challenged?

What to Practice Saying Out Loud

  1. A 30-second answer to any question in the bank
  2. A 2-minute answer with one example and one trade-off
  3. A follow-up answer that starts from assumptions