About Me

My path has traced an unusual arc — from theoretical physics, to the frontiers of large-scale industrial AI, and now back to the university to train the next generation of researchers. Few trajectories cross the academia–industry boundary once; even fewer cross it twice. I have spent the first half of my career proving research ideas in the lab, and the second half proving them in production at planetary scale. I have now returned to academia to bring both halves together.

I earned my BS in Physics from Tsinghua University in 2010, completed my PhD at Northwestern University in 2016, and spent a further year there as a Postdoctoral Researcher. My doctoral work reached both ends of the research spectrum — from the deeply theoretical, using Lyapunov functions to characterize stability in dynamical systems (published in Physical Review Letters), to the deeply empirical, modeling the dynamics of the entire U.S. power transmission grid (published in Science). That range shaped a research taste I still carry: I am drawn to work that either uncovers a fundamental principle or delivers impact at real-world scale — and, at its best, does both.

In 2017 I left academia and joined Facebook (now Meta), where I spent seven years inside one of the most demanding AI environments in the world, eventually rising to Senior Staff Research Scientist (E7). From 2023 to 2025, I served as the Tech Lead for Business AI Agent, driving it from zero to launch and putting it in the hands of millions of small businesses worldwide. Earlier, from 2018 to 2022, I led efforts on the core Ads machine learning team, developing privacy-preserving models that met strict constraints without compromising performance. My trajectory was featured by the SIAM community in 2022.

My 14 years in the United States can be summarized in a single equation: 7 years of academic training + 7 years at Meta Ads Core ML = a researcher with strong engineering instincts and an impact-driven mindset.

Now back in academia, I am focused on one of the most exciting frontiers in AI: building agentic systems that learn, collaborate, and act. I care deeply about clarity of ideas, strength of execution, and meaningful impact, and I am most energized by research that is not only publishable, but deployable — systems that stand up to the full complexity of the real world while advancing what we understand about it.

Lab Culture

Having spent most of my professional career at Meta (formerly Facebook), its culture has left a mark on how I think about research and collaboration. I strongly resonate with several core principles:

• Move Fast and Fail Fast: In a fast-moving AI landscape, quickly testing and discarding weak ideas accelerates learning and ensures we focus on what truly matters.
• Focus on Long-Term Impact: Research should be evaluated not only by publications, but by its potential to be adopted in real-world systems and create meaningful benefits.
• Build Awesome Things: I place strong emphasis on engineering quality and strive to build systems that are both elegant and practically useful, grounded in real needs.
• Be Direct and Respect Your Colleagues: I value clear, honest communication while maintaining mutual respect—this defines how I collaborate with others.