Building recommendation systems that daily recommends movies and books to more than 3 million users. Continuously improving these systems through AB testing.

Predicting how molecules move when exposed to lasers, with potential applications in controlling chemical reactions and materials science.

• Developed and implemented a new analytical theory for modeling types of laser-molecule interactions that was very difficult or impossible to model with previous methods. The new theory reduced computational time from months or years down to minutes.

• Implemented a highly efficient reference simulator in fortran to solve the numerically demanding laser-molecule interaction problem at intermediate laser parameters. This allowed me to validate the accuracy of the analytical theory.

• Built a research web page, using Hugo and Javascript, for organizing research notes and partial results, to overcome issues with my previous PDF based solution. Now my research notes can include dynamic content, be easily accessed from anywhere, and easily shared with collaborators.

• Launched a benchmarking effort of algorithms that are in use for solving an atom-laser interaction problem, in order to identify the fastest one. This involved coordinating between 9 research groups in the US, Europe and Asia. Found that the fastest method was more than 200 times faster than the slowest method.
• Built a python tool to manage and partly automate convergence testing, which is a tedious and time-consuming task. This made convergence testing much easier, and freed up time for other tasks.