Chemistry PhD · University of Chicago

I watch batteries while they work.

I point lasers into rechargeable zinc batteries to see what happens at the electrode as they charge and discharge. I design and 3D-print the cells that make that possible, and I’m working with machine-learning researchers on models that could help choose which experiment to run next.

Fig. 0 · SchematicEvery molecule scatters laser light into its own pattern of peaks, a fingerprint. This one is a map of the site: each peak is a page.

01 · Research

The main signal

All research →

Watching a zinc battery while it works

Most battery tests treat the cell as a black box: charge in, charge out, count what comes back. That tells you a battery is wearing out, not why. Raman spectroscopy reads the chemistry at the electrode surface as it happens, so a change in behavior can be traced to a change in molecules.

Methods: operando Raman spectroscopy, SHINERS, custom 3D-printed cells, Python analysis.

How it works →
How Raman scattering works (schematic)A green laser beam hits molecules on an electrode. Most of the light bounces back the same color. A tiny fraction comes back a slightly different color; that shift is the molecule's fingerprint, which a spectrometer reads.electrode surfacelaser inone pure colormost light: same color~1 in 10 million:a shifted color= the molecule's fingerprint
Schematic A tiny fraction of laser light comes back a different color. That shift is a molecular fingerprint.
Living page · Machine learning collaboration with Yuxin Chen's group

Teaching a model to pick my next experiment

Benchmarking simple and complex machine-learning models on my experimental data, then using active learning so the model helps choose which experiment to run next.In progress. Comparing ridge regression, random forests and neural networks; active learning is next.

02 · Projects

Things I’ve built

All projects →

Hardware & 3D Printing

3D-printed spectroelectrochemical cells

Custom PETG battery cells, designed so a Raman laser can look inside while they run, printed and iterated in-house on a Bambu Lab X2D.

  • 3D printing
  • instrumentation
  • CAD

ML & Pipelines

Spectral analysis in Python

The Python code that turns thousands of raw operando spectra into something a person, or a model, can reason about.

  • Python
  • Raman spectroscopy
  • data pipelines

03 · Now

What I’m up to

Now page →
  • Benchmarking ridge regression, random forests and neural networks on my experimental data with Yuxin Chen’s group, with active learning next.
  • Learning ML pipeline engineering and lab automation, in public.
  • Writing up my thesis research. PhD expected June 2027.

Updated 2026-10-30