Information retrieval · RAG evaluation · reproducible ML

Gioia Zheng

Information Retrieval · RAG Evaluation · Grounding · Reproducible ML

I study where retrieval-augmented generation fails, how retrieval and generation errors interact, and how evaluation can make those failures measurable. I build reproducible systems to turn these questions into testable evidence.

Featured research projectAll projects
rag-observatory
active

Research prototype for trace-based RAG observability and failure analysis.

ragobservabilityfailure-analysisreproducibility
Real traces40 SciFact
Controlled queries20 fixed
Automated tests107
updated 2026-07-27·View project
Recent writingAll notes
Why Better Retrieval Does Not Always Improve Grounding
2026-06-23

Better ranked evidence is useful, but it is not the same as a faithful answer. I treat retrieval quality and grounding as separate axes.

Read note