Joshua Daniell's current research project is an effort to provide probabilistic forecasts of solar indices/proxies, which are used as inputs (drivers) to thermosphere density models and directly affect the drag force modeling of objects in LEO. The goal of this work is to improve on the currently used short-term forecasting methods, provide a probabilistic forecast for the drivers, and provide a robust and reliable uncertainty estimate in the given forecasts. In order to generate these forecasts, his work utilizes machine learning (ML) methods, specifically neural network (NN) ensembles using model types such as Multi-Layer Perceptron (MLP) and Long-Short Term Memory (LSTM). By combining multiple neural network models via ensembling (similar to approaches for terrestrial weather), a better generalization of forecasted values can be provided. His work has generated improvements over the currently used forecasting method for the F10.7.cm solar radio flux proxy by leveraging the power of NN ensembles. He is currently working on similar approaches to generate improved forecasts for other solar indices/proxies, such as S10, M10, and Y10.