Peixin Wang

Self-improving search agents thrive when feedback and policy evolution are intertwined, leading to sustained performance gains and reduced hallucinations.

Accurately predicting distributed system performance under diverse fault conditions is now possible directly from formal designs, thanks to a new automated framework.