In her own words
What gets lost in translation can change a life.
I first understood the cost of missing information while interpreting Bengali in medical settings. A patient could describe pain precisely—its timing, texture, triggers, and consequences. Yet by the time the visit became a chart, much of that detail could be compressed into a few words. The meaning had been present. The record had not been built to hold all of it.
Years later, I recognized the same pattern in single-cell analysis. Rare cell states can contain the biology that distinguishes one patient’s disease from another, but integration and averaging can smooth those states into invisibility. The experiment may be technically successful while the signal that matters most is lost.
That parallel became the question behind my work: how can we build research and clinical systems that are sensitive enough to preserve what matters?
At ASU Biodesign, I began with models of breast-cancer recurrence. At Dana-Farber, I studied how ZAP-70 mutations change CAR-T cell function. At Stanford Medicine, I developed GenoRefine, a metric-guided framework for integrated single-cell embeddings. At Pfizer, I analyzed clinical-trial data and built tools linking dosing, laboratory findings, comorbidities, and clinical events. At Mayo Clinic, I collect tissue in the operating room and study how immune architecture, nerve signaling, and spatial tumor evolution shape disease.
The methods change. The responsibility does not: do not let the average become an excuse for overlooking the patient, the rare cell, or the community that does not fit the default.
I want my science to begin with listening, survive the translation into data, and return to medicine as evidence a clinician can act on.
