My dad is a doctor. For as long as I can remember, he sat across the dinner table and talked about his research, his institute, what it actually means to make a difference, and the oldest rule in his field: do no harm.
Those conversations shaped me long before I understood why. They shaped what I thought a meaningful life looked like, and for a long time I assumed mine would look like his. I wanted to go into medicine.
At some point (likely after I passed out from donating blood a few times too many) it started to feel like a door wasn't going to open for me in healthcare.
So I went another way. I studied mechanical engineering, and I ended up in Big Tech, where I spent a few years learning how much of the world runs on layers and layers of systems, coordination, and bureaucracy, and how hard it can be to see your own impact when you are one small part of something enormous.
Eventually, I got restless. My brain wanted shorter feedback loops and a closer line between the work I did and the impact it had.
So I left Big Tech and joined Fluidstack as a founding PM, building the software layer for a multi-gigawatt AI infrastructure buildout.
I got to work at the frontier, close to the labs and the physical systems that make this technology run. I led two products that became part of Fluidstack’s internal operating system for multi-gigawatt buildouts, spanning site development, hardware deployment, cluster readiness, and the first workloads on a live cluster.
I’m proud of that work. But the part that kept pulling at me was never the compute itself. It was where compute becomes something else.
I once tried to put "where compute becomes" on a hat for a partnership. It got cut. But the phrase stayed with me, because the unfinished part is the whole point.
Where compute becomes purpose. Where it becomes a sixteen-year-old feeling a little less alone. Where it becomes twenty minutes given back to a single mother, the lists and the meals handled, so that time goes to her kid instead. Where it becomes a step toward a cure. Where it becomes a real change in how we teach people. That in-between space, where infrastructure turns into something that changes a life, is where I want to pay attention.
Here's what I believe. We are compute-constrained. No question, we already are.
But compute is not the only bottleneck.
The harder constraint is knowing how to make increasingly capable systems safe and building the research, evaluations, policy, and product safeguards to back that claim up. Capability cannot outrun safety. If it does, there is a version of the future where all the compute in the world stops mattering.
Which is how I ended up where I am now: on OpenAI’s Safety Systems team, working across safety measurement, model policy, and product for the people who need safeguards most.
It turns out the oldest rule from my dad's dinner table is the one I care about most. Do no harm. I just found a different frontier where I can try to live it.
Excited doesn't cover it. I'm humbled, a little scared, and very ready. Let's see what compute becomes.