Co-founder
of Anglera.

−140 m · sunlight zone · Stanford

Learning how machines read the world.

BS and MS in computer science at Stanford, focused on AI. Up here the water is bright and everything seems legible — you train a model, it learns, the future feels close enough to touch. The lesson that stuck came later: a model is only as good as the data underneath it, and most of the world's data was never written for a machine to read.

Nobody tells you how deep the water goes. You find out by falling in.

−600 m · twilight zone · Uber Eats

Millions of SKUs, and every one a mess.

Product at Uber Eats: CPG ads for grocery, same-day delivery routing, the UberTasks pilot. Grocery catalogs flowed through those systems by the million, and somewhere between the ads and the routing we started noticing what we were actually standing on — the product data underneath commerce is a mess. Duplicate items. Missing sizes. Descriptions written for no one. We shipped feature after feature and felt the ground shift under every one.

We filed it under "marketplace problems." That was wrong, and being wrong about it changed our lives.

Millions of products swam past us every day. We couldn't stop seeing them.

−1,800 m · midnight zone · the realization

What isn't structured is hidden forever.

Look around. Every retailer. Every brand. Every distributor, every manufacturer — anyone, anywhere, with a catalog of SKUs is sitting on shelves of products the world can't read. Products without structure can't be understood. Products that can't be understood can't be found. Not buried. Not hard to find. Hidden forever.

Sit with what that means: a shopper wants a product. The product exists — in stock, priced, ready, perfect for them. They never meet. Multiply that missed meeting by billions and it stops being a data problem. It's one of the largest market inefficiencies left on the internet.

And the water was getting darker. Discovery was moving from blue links to AI chat, curated recommendations, direct checkout — agents that read structure, or read nothing. Same product: one gets recommended, the other doesn't exist.

Somewhere out there, someone needs exactly this thing. And they will never find it.

−3,200 m · abyssal zone · Anglera

So we brought our own light.

At this depth there is no sun. There never will be. And yet the abyss is not empty — because one small fish refused to accept the dark. The anglerfish doesn't wait for light to arrive. It grows its own, carries it out in front, and everything it needs comes swimming toward it.

That felt like instructions. So we did the unreasonable thing founders do: we went toward the problem everyone else had learned to live with, and we started Anglera — named for the fish. AI that does the enrichment work itself: sourcing, structuring, and scoring product data until an entire catalog is readable by the machines doing the world's shopping. Twenty years of tools promised to fix product data by storing it and moving it around. Anglera is the first that does the work.

You don't wait for the sun to reach the bottom of the ocean. You become the light.