I'm an ML engineer focused on multimodal AI for video understanding. I work end-to-end: from curating datasets and fine-tuning models to shipping production systems. My work has powered experiences for HBO Max, Google/YouTube, and most recently at OZU, where I lead ML as a co-founder.

The thread through all my work is teaching machines to understand visual storytelling. This includes elements of cinematography, narrative structure, plot understanding, emotional subtext, etc - the things that filmmakers know how to manipulate expertly but that's hard for viewers to articulate, or even detect without careful, deliberate consideration.

This obsession traces back to a film class in college where I spent hours analyzing a single 3 minute scene from Pulp Fiction. In 2019, I started teaching myself to code, wondering if machines could help with that kind of analysis, and have been building towards that ever since.

In a past life, I competed on the junior tennis circuit (world top 2000), acted in a couple of short films (Crumpled, Normal), and composed the soundtrack for one of them.