Starting the PhD Journey: From Xiamen to HKUST

I'm writing this from the tail end of my Master's at Xiamen University, with an offer to begin a Ph.D. in Chemical and Biological Engineering at HKUST later this year. It feels like a good moment to write down what got me here.

Finding the intersection

I came into graduate school with a background in analytical chemistry, not machine learning. The path from titrations and spectroscopy to neural network potentials wasn't obvious at the start — it took a lot of slow, deliberate relearning of the basics: linear algebra, optimization, the mechanics of how a model actually learns a potential energy surface rather than just memorizing training data.

What kept me going was realizing that the two fields ask the same underlying question in different languages. Quantum chemistry asks: given a molecular geometry, what is its energy? Machine learning asks: given enough examples, can a model learn the function that answers that question, fast enough to be useful at scale? Once I saw it that way, AI for chemistry stopped feeling like two separate disciplines bolted together.

What I'm taking with me

A few things I'll carry into the Ph.D.:

• A working knowledge of MLatom and the AIQM family of methods, built from the ground up rather than treated as a black box.
• A healthy respect for benchmarking — a method that looks good on one molecule and falls apart on the next isn't a method yet.
• The understanding that infrastructure work (data pipelines, reproducible workflows, careful version control) is not separate from "real research" — it's what makes real research possible.

What's ahead

HKUST's Chemical and Biological Engineering program gives me room to push further into the engineering side of scientific machine learning: building models and tools that other researchers can actually rely on, not just methods that work in a single paper's supplementary information.

More on that as it unfolds.