RESEARCH
I really like chasing questions that sit between different fields—art and religion, finance and engineering, or philosophy and AI.
① Dunhuang
For four years, I've studied the Dunhuang murals through three lenses: faith and color, food, and music. I’ve put everything I’ve learned into an interactive website.
How does an imported religion learn to look local? I spent a lot of time tracing Buddhist sculptures at Dunhuang from the Sixteen Kingdoms all the way to the Tang Dynasty. It was fascinating to see how the figures slowly took on Chinese facial features, body proportions, and clothing. I spent hours decoding the hidden meanings behind different mudras (hand gestures), ritual objects, and the animals the gods rode. I also looked into how natural mineral pigments—like cinnabar and lapis lazuli—were blended with shading techniques to make people feel like they were standing in a truly sacred space. This part of my project eventually led me to build a 3D reconstruction of Cave 217 and write a deep dive into the Contemplation Sutra mural on the north wall of Cave 217.
What did normal people in Tang-dynasty Dunhuang actually eat? To answer this, I paired mural scenes of grand banquets, street taverns, and religious offerings with actual ancient scraps of paper found in the Library Cave. I managed to piece together a picture of a very multicultural diet: lots of wheat-flour foods, exotic imported spices, and dairy and meat processing brought in by nomadic cultures. Official history books rarely tell you what people had for lunch. Looking at these murals and manuscripts side-by-side helped me fill in those blanks, showing how food wasn't just fuel—it was a huge part of people’s faith, identity, and community.
I also tried recreating one of these foods myself: sanzi, a fried twisted-dough snack, pulled by hand into thin, noodle-like strands.
I went through the murals and analyzed over 100 images of musical instruments, covering more than 40 different types painted over a thousand-year span. I traced how foreign arrivals—like the bent-neck pipa, the konghou, and the bili—slowly became localized right alongside native Chinese instruments like the ruanxian and sheng. It shows a really cool shift in ancient Chinese music from simple to incredibly complex, and from purely functional to deeply artistic. My absolute favorite part was studying the "self-sounding" instruments that float through the painted skies playing themselves. Even though they're religious symbols, they actually preserve highly accurate details of how musicians played back then, which is super useful for music history and digital heritage restoration.
These three strands came together in an essay I wrote titled "The Sacred in Small Things: Material Religion in Dunhuang Murals and The Canterbury Tales" (which I submitted to the Immerse Essay Competition). I also wanted to look at Dunhuang's modern-day challenges, so I started a project on tourist feedback—you can check out "Visitor Voices" on the Community page..
② Financing
How should a robotics startup pay for its R&D when it’s short on cash and the global supply chain might snap at any moment? In my paper, "The Impact of Venture Capital, R&D Investment, and Supply Disruption on Robotics Company Financing," I used math to model a supply chain made of one manufacturer and one unreliable supplier. Then, I calculated and compared the financial returns (NPV) of four different ways to fund the project: using your own cash, selling equity, taking out debt, or relying on customer pre-sales.
I found that using your own money is always the safest baseline. Selling equity lowers the final payout for everyone but pushes the risk onto investors. Debt only makes sense if you're sure demand will be huge or if the loan is tiny. Pre-sales, interestingly, can act as a great substitute for both equity and debt if you set them up right. Instead of just picking one "perfect" strategy, my model maps out exactly when a company should swap one option for another. This paper ended up winning a Regional 2nd Prize at the S.-T. Yau Science Award.
Embodied AI (things like smart robots, self-driving cars, and drones) is booming in China right now, but it takes a massive amount of capital. To make my previous model realistic enough for this industry, I completely rebuilt it. I added a backup emergency supplier, turned supply disruptions into random unexpected events, and factored in government R&D subsidies, investor incentives, and the real-world costs of running AI models (like servers and API tokens).
Then I asked a new question: which funding strategy wins as the market grows? It turns out that in small markets, the lines on the graph constantly cross—no single option is best. But as the market expands, funding pressures ease up and a clear ranking emerges. Investing early in R&D makes the supply chain way more resilient, and government subsidies take a huge weight off the firm's shoulders. This expanded project was selected as an ICBS Poster Session Nominee, making me one of just five students chosen from my school.
Both of these projects point toward the same open question, and it’s exactly what I want to solve next: what happens when a company doesn't just pick one strategy, but mixes a few of them together at the same time?
③ Zen × AI
Can the logic of a Zen koan—those paradoxical riddles designed to break you out of regular, rigid thinking patterns—actually help us write better prompts for AI? I've been exploring this idea under the guidance of Dr. Advait Sarkar from Microsoft Research (who is also an Affiliated Lecturer at the University of Cambridge). We're finding that by using Zen-inspired prompting structures, we can push AI models to be much better at helping humans solve open-ended problems and leveling up their own critical thinking skills.
④ Wharton
I was selected to join a summer cohort of about 75 students. We dove deep into machine learning, statistical modeling, and data visualization. For my final capstone project, I built a model that scrapes and analyzes social media sentiment to see how much public online chatter correlates with sudden stock price volatility.