Q&A
Welcome Houston. So how did Hubify begin its journey?
I'm not a physicist. I'm a software engineer who's been obsessed with astronomy and cosmology my whole life, but I never had the credentials to do real research. So I built the lab I wished existed — and ran it as one person.
How did you know where to begin?
I didn't wait until I had the perfect platform. I had a question worth answering, agents that could work around the clock, and a weekend. I set to the task of running real experiments on real survey data as quickly as possible.
So what was the first experiment?
One weekend and $200 of GPU compute over every public DESI spectrum — 17.65 million of them. That run became BigBounce, a six-paper cosmology program that every part of Hubify has been dogfooded on since.
What was the initial reaction?
Skepticism — which is exactly the point. Every result gets handed to reviewers from five rival AI labs whose only job is to tear it apart. The review record is public, round by round, and it made the work sharper every single week.
Where did you evolve from there?
268,519 validated anomalies across four surveys, an 8.47-million-galaxy chirality catalog, and six papers standing at the submission gate. The lab site publishes everything: the data, the reviews, the failures too.
Why adversarial review instead of agents that agree?
Cooperative agents polish each other's mistakes. Rivals hunt for them. When a claim survives five labs trying to kill it, you can trust it — and in science, what survives is the only thing worth publishing.