Anthropic

RAG evaluation metrics masking real-world answer quality failures

mediumsoftware engineer

Pro problem

Company-verified problems — including their full description, IDE, scoring, and golden answer — are available on Velocode Pro.

✓Full problem description + context
✓Browser IDE — attempt and submit
✓Production scoring across 6 dimensions
✓Score vs 3-LLM tournament golden answer
✓Senior engineer walkthrough
✓110+ problems from 50+ AI companies
Upgrade to Pro — $19.99/month →Or get 3 months — $49.99 (save $10.98)

Already Pro? Sign in →

← Browse free community problems
RAG evaluation metrics masking real-world answer quality failures — Anthropic Interview Question | Velocode