Jingxiaozhi AI Customer Service System
QA automation and NLP feature delivery for JD.com's e-commerce assistant
As a 2020 software engineering intern at JD.com’s AI Research Center, I joined two release cycles (2.1.6 and 2.1.7) to boost quality and analytics for the Jingxiaozhi conversational agent.
Highlights
- Built Java-based automated testing modules on top of JD’s JSF RPC framework to mimic real customer journeys, achieving 100% coverage for core API endpoints.
- Instrumented Redis- and RabbitMQ-backed telemetry to surface regression signals before feature flag rollouts.
- Partnered with the NLP team to evaluate Chinese-language intent models across retail scenarios and feed findings back into the training backlog.