I work on the systems that decide what content gets surfaced on Spotlight and Snap's main feed. Most of my work lives in the ML retrieval and infrastructure layer — building pipelines, modernizing inference systems, and migrating legacy infrastructure to more scalable solutions. Some of the highlights: cutting over $5M in annual infrastructure costs, improving content freshness metrics. I also jump in for oncall when things break, and I'm always looking for ways to use AI tools to work faster and smarter.

I worked on the ads side of Twitch, helping build out measurement tools and improve the reliability of the ads pipeline. I was part of the team that launched Twitch Brand Lift — handling the backend integration with Amazon's distributed services. I also built out the observability layer for the Audience Generation pipeline so the team could actually see what was going on when things went wrong. On top of that, I led backend work on an ad blocker mitigation project, working closely with Data Science to measure and validate the impact.