I support a team focused on applying machine learning to build advertising products that make better connections between people and advertisers. Day to day, our work involves applying a variety of modeling and AI techniques to get the most out of noisier, sparser and increasingly diverse data sets. As a big believer that many of the bottlenecks in real-world machine learning come from the gap between ideas and meaningful experiments, I also have interests in machine learning developer efficiency, experimental best practices, and am involved in machine learning education at the company.
Before Facebook, I did my postdoc at Stanford, where my focus was on developing data analysis and modeling methods to help interpret X-ray diffraction patterns coming from the free electron laser at SLAC. Before that, I did my PhD and undergrad at Oxford university, where I worked on modeling the propagation of X-ray damage through biological crystals. Surprisingly, my day-to-day life at Facebook is not very different from my academic work: hypothesis — model — experiment — repeat!


Ads, machine learning, sparse modeling, learning developer efficiency, experimentation, and neural networks