Design RAG Evaluation and Debugging
Company: Plaid
Role: Machine Learning Engineer
Category: ML System Design
Difficulty: medium
Interview Round: Onsite
Overview: This question evaluates a candidate's ability to design evaluation and debugging frameworks for production RAG systems, testing practical ML engineering skills in retrieval-augmented generation pipelines. It assesses system design competency around offline and online quality metrics, regression detection, and failure localization in high-stakes domains requiring strict data permissions and faithfulness guarantees.