Build evals for LLM-driven applications and iterate on them with measurable progress. Covers eval set construction, automatic + LLM-judge scoring, regression testing, and how to actually use evals during development. Use whenever you're building anything that calls an LLM in production.