I work on one problem: read what a person actually means, put the right option in front of them, and prove the choice was right.
Most people assume that work is understanding the person: intent inference, the goal under the request. That is only half of it. The other half is understanding the options, and it’s the half that goes unbuilt. Both sides have to be written into one shared ontology, in the same terms, because two rich models in different languages connect no better than two shallow ones.
I’ve built that connection in more than one industry. In food service it was 400,000 restaurants on one side and hundreds of thousands of suppliers on the other. In commerce it’s shoppers and ecommerce catalogs, which arrive thin and messy: a title, a few attributes, a guess. At Velou we describe each individual product into one shared ontology, millions of them across catalogs, deep enough that a shopper’s intent can actually find the right one. Different domains, same machine: products, suppliers, flights for a family’s week away, the meetings on a calendar. Solving this problem is my life’s work.
The rest is connection: rank the intent to the right option, act on it, then measure what happened and feed it back so the loop actually improves.
I’m the founder of Velou; what I think about all of it lives in the writing.
Born in Sri Lanka, grew up in Australia, now home in New York with my wife, Alex. A few times a year I’m back in Melbourne with my nieces, who are reliably unimpressed by any of it. I box, less for the sport than the precision of it.
If you want the ideas, start with the writing. To get in touch, see Contact.