Xiao is currently a software engineer at Google Deepmind (2024 - present), with extensive hands-on post-training experience in Gemini safety evaluation and mitigation. Most recently, she has focused on developing new post-training modeling techniques that leverage the model's reasoning ability to produce reward, enabling self-improve/align through Reinforcement Learning with applications in Gemini safety and beyond.

Xiao's prior research focuses on understanding how AI increasingly mediates social exchange and its societal impact on trust. Previously, Xiao worked at Facebook Core Data Science (2018) on trust in social groups and Airbnb (2016) on trust in host profiles.