At Lyft, I worked on buidling tools for scientists, including simulation. I also worked on many interesting optimization, forecasting, and machine learning problems.
For my PhD research I wrote software to schedule electrical power systems with uncertain wind energy. I focused on improving algorithms to create wind power scenarios from forecasts and historical data. I evaluated these algorithms based on stochastic unit commitment simulations of the Texas power system. Each simulation was a series of large stochastic optimization problems, which I ran on the UW's high-performance computer cluster. Along the way I created a Python package for power systems optimization called Minpower.