LLM Inference Interview Questions, issue 20, Aug 20, 2026
The Diversity Collapse Trap
Staff ML Engineer interview at Google DeepMind, and the interviewer asks:
“AlphaCode picked submissions by clustering programs on behavioral equivalence. AlphaCode 2 used a fine-tuned scoring model instead. Your junior wants to rip out the clustering and ship the reward model tomorrow. What breaks?”
Don’t say: “Nothing, a learned reward model is strictly better than a heuristic.”
Why replacing heuristic clustering with a pure reward model silently hands you 10 copies of the same wrong answer, and how to stack them to balance precision with coverage.
The full answer, with the mechanism and the arithmetic, is for paid subscribers on Substack.