RAG Interview Questions, issue 18, Jul 22, 2026

The Hop-Count Paradox

Senior ML Engineer interview at Google, and the interviewer asks:

Adaptive RAG routes queries by predicted hop count, but nobody labels this question needs 2 hops. The paper bootstraps silver labels by brute-forcing every hop count and keeping whichever one got the right answer. What’s the ceiling that creates?

Don’t say: Silver labels are noisy, so you need more data cleaning.

When predicting query complexity upfront silently memorizes your pipeline's weaknesses, and why treating retrieval as an evidence-gated policy prevents massive latency waste.

The full answer, with the mechanism and the arithmetic, is for paid subscribers on Substack.

Read it on Substack

Get the next one

Free on Substack. Unsubscribe in one click.