Tumor biology & quantitative imaging
I use imaging and molecular markers to target the biology inside a tumor.
Most recently, I worked with Dr. Howard Halpern at the University of Chicago on tumor hypoxia. We mapped molecular oxygen deep inside tumors with electron paramagnetic resonance imaging and asked how that map should change the way radiotherapy is delivered. The lab pointed that same oxygen-mapping approach at Alzheimer's disease and at fentanyl overdose and rescue too.
Before that, I worked on turning the immune system against the tumor microenvironment in Dr. Bin Zhang's lab at Northwestern, and on β-cell insulin resistance in zebrafish as an undergraduate at Vanderbilt.
Ran in vivo EPR, MRI, CT, and PET imaging on murine tumor models independently, along with radioisotope handling and radiation administration. When health difficulties left the lab short-staffed, I took over every animal experiment in the group for four months with no lapse in data quality or protocol compliance.
Flow cytometry and PCR-based genotyping across multiple cancer models, working on immune targeting of the tumor microenvironment. Independently genotyped several hundred mice and absorbed full colony management when a postdoc went on leave, keeping the research uninterrupted.
Studied β-cell insulin resistance using zebrafish. Ran gateway cloning to build genetic constructs, used CRISPR for gene insertion, and fluorescently tagged proteins to track them in real time — the project that taught me how to take an unfamiliar protocol to independent execution.
Teaching a fly's brain to play the drums
The idea starts with something the fly already does. A male hears another fly's wingbeat through Johnston's organ, and that signal routes through a single descending neuron, pIP10, into courtship song: one wing beating out a species-specific rhythm. flybeats hijacks that exact pathway. The input becomes a song instead of a rival's wingbeat, and the output becomes a drum kit instead of a courtship performance.
MaleCNS v1.0 is a full synaptic wiring diagram of a fruit fly brain: every neuron, every connection, released by Janelia and Google Research. flybeats keeps that wiring fixed and only lets the connection strengths change, then trains the network to listen to a song and drive a drum kit through the fly's own wing-steering motor neurons.
It's a strange question dressed up as a fun one: how much of "sounding musical" comes from the wiring itself, versus the training? The pipeline runs end to end: real-time playback, an adjustable tempo dial, even a lesion mode that mutes individual neuron types mid-performance.
View the code on GitHubI'm drawn to contrast — light against shadow, rust against glass, whatever doesn't quite belong sitting next to whatever does.
The other side of flybeats: I make and mix my own tracks, which is probably why teaching a fly to keep time felt like a natural side quest.
Long rides are where half of my best troubleshooting ideas show up uninvited.
The same process as the bench: read the method, change one variable, taste, repeat.
Loads eight photos from my portfolio whenever you refresh the page.