Meet the people behind KaraOrchee.

I was eleven when I left home to study at one of China's most prestigious music schools. Back then, my world was pretty straightforward: practice, lessons, competitions, repeat.

That routine eventually took me to Oberlin, onto concert stages, and into performances with symphony orchestras. I won the Wideman Gold Medal and the Naftzger Grand Prize, recorded my first album, and somehow ended up teaching the next generation of pianists, too.

But here's something people don't often see.

You can spend eight hours practicing a concerto in a room by yourself. And when you finally get to rehearse with an orchestra, you may discover that the orchestra has an entirely different opinion about your tempo.

I kept wondering: why, with all the technology we have, can't I practice with something that actually listens to me?

That question eventually became KaraOrchee.

At first, I wanted to build an orchestra that could follow a soloist. Then I started thinking about my students: the little discoveries we make during lessons, the things I explain three different ways until something finally clicks, and how much of that gets lost by the following week.

Now I spend a rather unusual amount of time discussing software, AI models, music notation, and product design with my co-founders.

I still practice. I still teach. And I still have a habit of asking for features that Justin tells me are considerably harder than they sound.

I usually ask again anyway.

I've spent more than twenty years with music, studying both cello and composition.

Playing and writing music are two different experiences. When I perform, I think about the phrase, the sound, the person sitting next to me. When I compose, I have to decide exactly how to communicate those things on a page.

And then there's the question of whether anyone will interpret that page the way I intended.

Now imagine asking a computer to do it.

A missing slur can change a phrase. A repeated section can send you back to the wrong measure. A rest that looks unimportant might be the very thing holding an entire passage together.

These are the kinds of details I find myself thinking about at KaraOrchee.

I work on our repertoire, music-library production, score accuracy, and operations. That means moving between Beethoven sonatas, MusicXML files, technical discussions, and the occasional very long conversation about notation.

It's an unusual combination of jobs.

But when you're trying to teach a computer something musicians have spent centuries learning, the details aren't really optional.