The Arc
The path from biomedical optics to building AI systems that see.
The Bowden Lab
I spent two years trying to help surgeons see cancer. In a biomedical optics lab at Clemson, I worked on removing specular reflections from endoscopy video — the bright spots that obscure tissue and make it harder to identify tumors in real time.
That work taught me something fundamental about machine perception: the hardest part isn't the model. It's the gap between what a sensor captures and what a human needs to understand. I published methods for flow-guided video restoration, but the question that stayed with me was bigger.
How do we teach machines to see the way biological systems do?
Multimodal Fusion
What if different sensors see the same thing differently — and that's a feature, not a bug? My research into vehicle re-identification and multimodal fusion at NeurIPS led me to UniCat, a unified approach to cross-modal attention.
The realization was that most fusion architectures were broken in the same way: they treated modalities as equal channels to be merged, rather than as complementary perspectives to be triangulated. Biological perception doesn't fuse — it triangulates.
This insight shaped everything that came after. Intelligence isn't about having more data. It's about having the right relationship between different kinds of data.
Data at Scale
The model is only as good as what you feed it. At Datology, I learned this lesson at scale — building the infrastructure for CLIP pretraining, curation pipelines, and evaluation systems that actually change the next training decision.
I turned month-long research cycles into weekend-scale iteration loops. I accelerated VLM evaluation by an order of magnitude. But the work I'm proudest of isn't any single model.
It's the infrastructure that lets a research team move faster and trust their results more. The systems that make the next experiment possible.
What's Next
I'm looking for the question I haven't asked yet. Robotics and embodied AI are pulling me forward — what happens when intelligence gets a body? When perception has consequences?
I recently picked up viola again after years away. I'm rethinking my writing toward the things that actually fascinate me: applied AI through a biological lens, the philosophy of perception, the spaces between disciplines.
Maybe the purpose of life is to just be alive. I'm trying to take that seriously — to build a life that's as intentional as the systems I build at work.