The Story
I've always wanted to understand how things actually work. In college I studied genetics because I was curious about the machinery inside cells. I ended up doing research on virtual screening—testing millions of drug candidates on a computer instead of in a lab, and I thought, well, this is interesting. Then I got curious about something else: how does business work? So I spent a few years doing consulting, delivering custom processes and software for pharmas, life science, supply chain, banks, and medical device companies. What I found was that I really enjoyed the part where you figure out what to build and then watch people use it.
That led me to product management, which I merged with my love of biology. At DNAnexus I led their precision medicine platform and built/led a scientific product R&D team. We made it possible for researchers to analyze huge genomic datasets without having to think about the infrastructure. It grew to be a large portion of the company's revenue and we also got to power the UK Biobank's research platform. At AWS I did similar things at larger scale: built a genomics platform, shipped drug discovery tools with AI integration, turned around a healthcare NLP product. The interesting pattern I noticed was that compute and data often weren't the bottlenecks. The bottleneck was that scientists didn't know what to ask for from software, only how to express the problems they wanted to solve. I also found that most AI systems people were building didn't know how to meet scientists where they're at.
So now I'm trying to understand that gap better. I've been building things from scratch: an agent skill library for bioinformatics (to see if agents can handle the nuance), a model that tries to predict how cells respond to perturbations (to see if we can simulate biology instead of just measuring it), and a consumer-focused AI trainer app (to learn what it takes to ship agents to real users). Each one teaches me something different about what works and what doesn't.