CrackML by @ml.with.umang
Interview questions / GenAI & LLM
GenAI & LLM interview question

Explain tokenization choices for an LLM

Explain tokenization choices for an LLM.

mediumconceptEvidence 40/1001 source reportAmazon

The 60-second answer

Tokenization maps raw text into discrete IDs consumed by the embedding table. Subword methods balance vocabulary size, sequence length, and out-of-vocabulary robustness.

Build the answer in this order

1
Define the mechanism

Tokenization maps raw text into discrete IDs consumed by the embedding table.

2
Explain the architecture

Subword methods balance vocabulary size, sequence length, and out-of-vocabulary robustness.

3
Compare trade-offs

Compare BPE, WordPiece, and unigram approaches conceptually rather than by brand name alone.

4
Close with serving + evaluation

Measure downstream quality and token-cost impact for the target languages/domains.

A useful interview mental model

This is the shape of a strong answer—not a script to memorize.

01Definition
02Mechanism
03Trade-offs
04Failure modes
05When to use

Senior-level signal

  • Tokenizer changes break checkpoint and embedding compatibility.
  • Multilingual and code-heavy workloads need special attention to fragmentation and normalization.

What the interviewer is really testing

Understanding beyond prompting: architecture, retrieval, evaluation, inference, safety, latency, cost, and failure recovery.

Likely follow-up questions

What assumption makes this approach work?
When would you choose the strongest alternative instead?
What production or data failure mode changes your answer?

Common weak-answer patterns

  • Reciting a definition without mechanism or assumptions.
  • Claiming one technique is always better without a data regime.
  • Stopping before failure modes, validation, or deployment implications.