September 2026 MMLI Community Spotlight

In August, Changhyun successfully defended his dissertation, “Autonomous Experimentation for AI-Driven Discovery of Functional Molecules,” earning his Ph.D. in Chemical and Biomolecular Engineering from the University of Illinois Urbana-Champaign. His graduate work in Professor Ying Diao’s group focused on building integrated systems connecting polymer thin-film processing, scientific instrumentation, experimental automation, and data-driven optimization. Changhyun’s doctoral research also included a study combining closed-loop experimentation with interpretable machine learning to uncover factors governing molecular photostability. This co-first-authored work, titled “Closed-loop transfer enables artificial intelligence to yield chemical knowledge,” was published in Nature in 2024.

Prior to joining NSF MMLI, Changhyun began exploring experimental platforms at KIST Europe, where he researched biosensors for heavy-metal detection. Since then, his interest has evolved into a broader curiosity about materials research. Specifically, he envisions systems connecting fabrication, characterization, data, and AI to learn from experimental results and guide what happens next. Even though Changhyun is entering a new phase of his career, his curiosity and vision will undoubtedly continue to drive his exploration of what becomes possible at the intersection of science, automation, and AI.

Since 2020.

Before beginning my Ph.D. at UIUC, I worked as a Research Assistant at KIST Europe in Saarbrücken, Germany. There, I developed a biosensor platform for heavy-metal detection by integrating nanoparticle–peptide chemistry with experimental assay development. The work gave me early experience in interdisciplinary research spanning materials, biology, chemistry, and sensing, and ultimately contributed to a peer-reviewed publication. That experience also helped shape my interest in building experimental platforms that combine materials design, measurement, and data-driven experimentation.

I lead the development of an autonomous additive manufacturing platform for organic thin-film electronics. My work focuses on integrating automated thin-film fabrication, multimodal characterization, data infrastructure, and Bayesian optimization into a closed-loop experimental workflow. Using this platform, we systematically explore processing conditions and uncover processing–structure–property relationships in conjugated polymer thin films, with the broader goal of accelerating materials discovery and process optimization.

I was drawn to MMLI because of its strongly interdisciplinary approach to materials research. I was particularly excited by the opportunity to work on AI-guided materials discovery alongside researchers with expertise spanning materials science, chemistry, automation, and machine learning. That combination of perspectives makes it possible to approach materials discovery in ways that would be difficult within a single discipline.

My favorite part of being involved with MMLI is the collaborative environment. The institute brings together researchers with very different technical backgrounds, which makes it easy to exchange ideas and look at the same scientific problem from multiple perspectives. I also really value the opportunity to work on projects that connect fundamental materials science with emerging tools in automation and machine learning.

One of the most fulfilling aspects of my experience with MMLI has been working closely with researchers across different disciplines within Thrust 4. Through collaborations and discussions with PIs including Professors Martin Burke, Nicholas Jackson, and Charles Schroeder, I have been exposed to very different ways of thinking about materials discovery, from chemistry and molecular design to polymer physics and data-driven experimentation. Learning from those perspectives has significantly broadened the way I approach research and has been an important part of my growth as a scientist.

I’m a big fan of crime, mystery, and investigative shows – especially the kind where you find yourself trying to solve the case before the characters do.

Slow Horses! It’s one of my current favorites – you should definitely check it out!

1. Spend a week surviving somewhere completely remote – maybe even on a deserted island. I’ve been preparing (theoretically, not physically!) for years by watching survival videos on YouTube, so I’m pretty much qualified at this point.
2. Drive a sports car along some of the world’s most scenic coastal roads – ideally with great scenery, low traffic, and no speed cameras.
3. Watch a UEFA Champions League final in person. Come on Arsenal!

I tend to treat vacation planning like an optimization problem. I often spend more time building the itinerary than actually traveling, so by the time I get there, the trip feels a little like validation – checking whether all the restaurants, routes, and viewpoints I planned actually exist in real life.

“Home” has become a little hard for me to define. I grew up in Korea, spent some time in Germany, and have now lived in Illinois for the past several years. At this point, home feels less like one particular place and more like wherever the people I care about are.

Our Lab’s official GitHub repo.

In our August episode of PEERspectives, a YouTube series exploring emerging research through literature reviews, Ziang was our guest host. He examined a Nature Methods publication demonstrating how co-distillation of Evolutionary Scale Models (ESMs) can improve variant-effect prediction (VEP). Meanwhile, Ziang recently developed SPURS, a deep-learning tool designed to help researchers understand, engineer, and evaluate proteins. The framework combines sequence- and structure-based protein generative models to predict how mutations affect protein stability. Published in Nature Communications in December 2025, the work complements his broader research into how proteins fold, move, and interact. His long-term ambition is to use machine learning to design molecular machines and programmable structures.

Whether it’s earning a gold medal in a Chemistry Olympiad in high school, advancing R&D in an industry lab, or building computational tools for protein research with MMLI, Ziang consistently develops interdisciplinary fluency throughout his academic career.

1.5 years

My background is in computer science. I received my undergraduate degree in Computer Science from Tsinghua University, and before that, I participated in Chemistry Olympiad competitions in high school.

I am interested in understanding how protein stability affects function, as well as how protein dynamics shape protein function. In MMLI, I mainly work on developing machine learning methods to study these relationships and ultimately use them for enzyme design.

I was drawn to MMLI because Prof. Luo is a great mentor, and I wanted the opportunity to validate my computational methods through wet-lab experiments and connect algorithm development more directly with real biological questions.

My favorite part is the community. I’ve made many friends through MMLI, and I really enjoy discussing new ideas with people, especially around enzymes and protein science.

The MMLI Retreat in 2025 was especially meaningful to me. Since I am not based at UIUC, it gave me a great opportunity to meet people across MMLI, get to know the broader community, and have many interesting research conversations in person.

Work out more.

Live in Yunnan after I retire. Design a molecular machine from scratch within the next 20 years. Get really into scuba diving.

My personal webpage.