The canon
I started piano at eight. My first teacher lived next door, and I learned without much rigour: my playing was unstable, unrefined and very dramatic. It must also have been musical, because I kept being passed on to more serious teachers. I finished the ARCT at thirteen and spent the next few years winning competitions, up to the Canadian Music Competition and the Seattle International.
At fifteen I joined a music academy where my fundamentals were torn down and rebuilt. I played Rachmaninoff's First and Prokofiev's Second concertos with orchestra, and learned Gaspard de la nuit, the Hammerklavier and Islamey. I never practised scales or Hanon, and I still don't.
Masterclasses with Kirill Gerstein and James Anagnoson.
Uncharted repertoire
University nearly ended it. Computer science at Waterloo took everything, and I was tired of the standard repertoire; I thought I had heard everything worth hearing. Then, at nineteen, I found Medtner's Sonata-Ballade: the first music in years that I understood, and that understood me.
Feinberg, Roslavets, Ives and Szymanowski followed. At twenty I started practising seriously again and recording for the channel. At twenty-one came Sorabji, the first music in a long time that was excruciating to learn: Gulistān, Djâmî, and then the four hours of Opus Clavicembalisticum.
I play Gulistān from Marc-André Hamelin's edition.
The in-between
My playing has changed more than my technique. At thirteen I knew how to make a piece sound difficult. At fifteen, how to make a moment memorable. At sixteen, how to make playing sound impersonal and effortless, as if nobody were at the keyboard. Sorabji taught me how to make sense of music that is abstract, complex and very long.
Now I try to play the in-between. If every moment is intense, intensity becomes a formula. I doubt the composers played their own music with that kind of personal immersion; they were illustrating a scene or an idea. When I go back to Schumann, it feels like reliving a memory for the first time, in sound.
Years of sight-reading became a method: a repeatable process in place of trial and error.
The reductions
In 2024 I turned to the operas and symphonies I loved most, beginning with Puccini, and started reducing them for solo piano. Since 2025 most of my time at the piano has gone into them: Strauss, Bruckner, Shostakovich. I don't simplify and I don't paraphrase. The aim is the smallest possible loss of musical information under one hard constraint: ten fingers.
Figurations stay figurations and tremolos stay tremolos. The vocal lines are folded into the texture instead of lifted above it. Then I play every reduction through, because a reduction you can't play isn't finished.
Published on IMSLP under CC BY-SA 4.0. Open on purpose.
The other hand
All of this happened alongside a computer science degree at Waterloo and six internships, then Amazon, then a startup where I led the research behind a patented video watermark, and now a principal role building AI systems.
The work is the same shape. Find the root cause. Respect the constraint. Lose as little as possible. Whether the source is an orchestral score or a live video stream, I'm compressing something large into something that still has to work.
