6 years
Building and leading the ML capability of the Vector Engineering Program at the Broad Institute from the ground up
Named inventor
Patent portfolio spanning ML-guided protein engineering, AAV capsid optimization, and CNS gene delivery
Nature Communications, Largest ML study for AAV capsid engineering
10+ years
Auditing bioML models, identifying why they fail before they reach the clinic
The field problem
No single lab or company will solve all of AAV's problems. I want to be part of moving the entire field, not optimizing within one program while the same structural problems repeat everywhere else.
The skills problem
I have spent a decade at the intersection of AAV biology and machine learning, and two decades in ML broadly, a combination that is rare in this field. Most programs will never have access to this kind of judgment. Working independently is how I make sure more of them do.
The training gap
>80% of the bioML models I audited failed to generalize; their authors never saw it coming. The problem was never talent. It was a training gap that standard ML education does not cover.
I work on what matters most.
I serve wider.
The science to help the patients who need gene therapy exists.
The barrier is the time and cost it takes to get there.
That is the gap I am working to close.
For the field
RAISE THE FLOOR
Share what works
Expose what fails
Build tools everyone can eventually access
For programs
RAISE THE CEILING
Build computational infrastructure
Move from ML-optimized to IND-ready
Inform decisions before they are expensive
No lab
I built it without institutional backing
No team
I designed and built every module
No funding