I am a research scientist at Facebook. My work centers on understanding the experiences and challenges of marginalized communities and designing and building inclusive and empowering technologies with the community.

My research interests include:

  1. AI for Empowerment & Equity: Design and develop AI products and experiences that center on the needs of the marginalized. Deliver tangible benefits and reduce the barrier for participation for those who had been under-served and oppressed.

    • AI & accessibility: How AI technology can be built to augment human abilities and empower people with different capacities to fully participate in and contribute to all aspects of our society without feeling constrained or stagmatized by one's physical/mental abilities and societal biases. And the challenges that come alongs.

    • Human-centered AI: How to engage and incentivize the deep learning community to practice and appreciate participatory design and community experience research in the development cycle of AI and stop equivalizing people as labels or embeddings.

  2. Computational Social Science: Understand and improve aspects of our socity - especially, the experiences of marginalized communities in our society, through qualitative and quantative research and computational methods? Most of the topics I have been working on - e.g. accessibility, gender inequality, racial justice, literacy - are complex societal challenges with deep sociocultural roots and I do not expect them to be solved by technology alone. However, I do believe that online activities and social media data can offer us new perspectives and understanding of these challenges that uncover opportunities where carefully designed technologies can make an meaningful impact. In particular, I have been and will be eager to explore:

    • How marginalized and minority groups form communities and build collective consciousness online.

    • Where are the gaps in online experiences and how they linked to gaps in offline outcomes for people from different social groups.

    • How to make informaiton and services accessible for all? Meeting the W3C accessibility standards is the first step but also a minimal requirement, we need to both empower under-priviledged groups with knowledge and resources but also design information systems that meet users with their indivisualized needs.

I founded the AI for Inclusion initiative inside the Society and AI Lab (SAIL) at Facebook AI, through which I build AI-powered product experiences to empower and include people from marginalized and under-resourced communities. One example of such experiences is automatic alt-text, a feature I designed and shipped on Facebook that leverages computer vision to describes images to people with visual impairments. More recently, I worked with people with dyslexia to build AI-powered writing tools to support them to write more confidently on social media.

Before joining Facebook AI, I was part of Facebook's Core Data Science team where I researched temporal and structural models to predict how information and behavior spread in social networks.

I am a big fan of mixed methodology and embrace both qualitative and quantitative research methods. My current research takes place at the intersection of AI and HCI, as I believe AI research should be user/problem oriented rather than technology oriented.

I publish my work at CHI, theWebConf (formerly known as WWW), CSCW, and ICWSM. I frequently review papers and serve in the organization committee for journals and conferences as well.

I received my PhD in Information Science from Cornell University, advised by Prof. Michael Macy, Prof. Jon Kleinberg, and Prof. Dan Cosley. I am also heavily influenced by Lada Adamic and Prof. Phoebe Sengers, botha world-class computer scientists but with very different research interests and methodology - I am grateful for their mentorship and teaching.

To learn more about our research and publications, you can also visit: