I'm Joe: a machine learning engineer and computational social scientist. Currently, I work at Replicate, where I focus on language models and ML infrastructure.

Before Transformers took over the world, I used ML and NLP to study phenomena like violent protests and acts of hate . I also worked on various aspects of NLP, like corpus development, language dynamics (e.g. here & here), and the measurement of complex semantic constructs (e.g. here & here).

However, after a postdoc at Northwestern, I decided to leave academia. I was drawn to the potential of LMs and I wanted to help build the future.

Since then, I've worked with language models in many contexts, including R&D, applied science, product development, and ML engineering.