The University of Arizona Power Forecasting Group uses scientific Python to research and produce solar and wind power forecasts for electric utilities in the Southwest. We will describe our forecast generation system that relies on Dask.distributed for easy parallelism and quick transitions from research to operations. We will also describe three Bokeh applications we’ve developed to explore the output of numerical weather models. Two of these applications can be viewed at https://forecasting.energy.arizona.edu/advi and https://forecasting.energy.arizona.edu/artsy.