I'm a ML researcher & systems engineer based in San Francisco. I build platforms and infrastructure with a focus on language modeling, web frameworks, and developer tooling.
Recent Writing
Work & Projects
I'm currently building MonkeySee to make it easier to test and automate webapps. We're using the latest language modeling and computer vision to understand HTML and act on it. I'm also a co-founder at Saywhat alongside Will McTighe, where we're making it easier to build businesses through the power of social networking.
On the side I maintain Mountaineer, a Python and React framework for building full-stack web applications with type safety, and Envelope, a private Python package registry for teams.
Before this I spent just over five years at Globality, starting as a Software Engineer and eventually becoming Vice President of Machine Learning. I got a chance to grow a big team, train language models from scratch, crawl the web, and build a whole inference stack in the days before vLLM.
I studied at Stanford University where my core research was on multimodal models for language modeling and vision understanding. I worked a lot with early transformers. During that time I was also an Engineer in Residence at Founders Circle Capital.
These days I make small seed-sized investments in promising deep tech startups and advise companies on engineering architecture and AI strategy. I also record conversations about AI and engineering over at pretrained.fm.