Design a feedback-text personalization data system
Company: Ancestry
Role: Software Engineer
Category: ML System Design
Difficulty: medium
Interview Round: Technical Screen
## System/ML design scenario: personalize using user feedback text
You are building an embeddings + personalization pipeline for a consumer product (e.g., genealogy/ancestry). Users provide **free-text feedback** (comments, reviews, support tickets) about products/features.
Design a system that:
1. **Ingests and stores** user feedback text so it can be queried and used for modeling.
2. Defines **features/representations** (e.g., embeddings) to find **similar users and/or similar products** from feedback.
3. Supports **automatic updates of “input categories”** (e.g., tags/topics/issue categories) as new kinds of feedback appear.
Assume moderate scale (millions of users, tens of millions of feedback records). Discuss data model, pipelines, model choices, retrieval, updates, and evaluation.
Quick Answer: This question evaluates skills in designing end-to-end ML systems for personalization from free-text feedback, covering competencies in NLP feature representation (embeddings), data modeling, retrieval, pipeline orchestration, model selection, and scalability within the ML System Design domain.