Xueying Liu

Problem: Customer requests were fragmented across multiple support scenarios.

Role: Designed the workflow, service logic, and response structure.

Approach: Mapped user intents and built a multi-agent flow with scenario-based routing.

Output: Created a clearer and more scalable support structure for relevant responses.

Problem: Raw user comments were noisy and difficult to interpret at scale.

Role: Designed the analysis logic and insight extraction process.

Approach: Combined LLM API, NLP preprocessing, sentiment classification, and topic clustering.

Output: Transformed user comments into structured signals for content and product understanding.

Problem: Customer behaviors were not clearly prioritized for retention action.

Role: Conducted segmentation and customer value modeling.

Approach: Used RFM analysis, BG/NBD, Gamma-Gamma, and CLV estimation.

Output: Identified priority customer segments and supported differentiated retention strategy.