Abdullah Nauman

I'm interested in the future of programming, specifically agent harnesses, orchestration, and human collaboration.

I'm the co-founder of zag.dev - the modern GitHub alternative for coding agents. We're rethinking code review and source control to enable orders of magnitude more agent-driven changes per day. We provision dedicated clusters for large enterprises and are backed by YC, Lightspeed, and Susa Ventures.

Previously, I was at Google Search, designing cool AI features for the results page that anticipate user intent.

Before that, I built ML infrastructure for conversion modeling, pioneering a DNN-based approach to joining billions of unattributed clicks across the web. Our 4-person team drove $1B in revenue for Google in 2024.

I studied at Michigan and took a gap year to join Agapè as employee #2. We built a viral growth loop that scaled the product to 200k+ users, caught Michael Seibel's attention, and got us into YC S21.

Earlier, I developed a distributed compression algorithm for imaging silicon wafers at KLA and worked on designing Pharos at JSTOR.org.

I also researched high-energy physics models at CERN and BNL with Abid Malik, optimizing distributed training on Summit, then the world's fastest supercomputer. I presented at Supercomputing '19.