I love trailblazing into the future, and improving the way artificial intelligence will shape our digital lives. In particular, I believe in improving the interplay between human cognition and computer instruction and in making both directions effortlessly smooth by crafting tools using using both deductive (static or dymanic) and intuitive (ML) reasoning.
I started at GitHub in 2019, when it acquired the code intelligence company Semmle, where I was part of the Data Science team. I've since been a part of GitHub's ML-on-code group, until I joined Github Next to be one of the starting trio exploring the novel code synthesis techniques that we turned into GitHub Copilot. Within that project, I was most deeply involved in promptcrafting, experimentation, and post-processing.
Before joining GitHub, I worked on developer productivity and on augmenting static program analysis with neural learning at Semmle, and before that on various ML applications to pharmacological and food science topics for Tessella. My academic background is in Logic and Proof Theory, where I wrote my PhD about constructive set theory.