Aparna Dhinakaran

Background

Co-founder at Arize AI, which builds LLM observability and evaluation platforms. Author on LLM-as-a-judge methodology.

Articles in This Vault

• LLM-as-a-Judge When to Use Reasoning CoT and Explanations — practical guide on explanation ordering, CoT tradeoffs, and reasoning models for evaluation

Key Contributions

• Evidence-based guidance on when explanations improve LLM judge reliability
• Practical prompting patterns: explanation-first, label-only, structured CoT
• Analysis of reasoning model tradeoffs (cost vs. quality)

Concepts

• LLM as Judge
• Search Evaluation