Our group addresses this problem by investigating the magnetic activity of the Sun and using it as a calibrator in the search for stellar magnetic twins. Spatially resolved solar observations let us establish how magnetic processes map onto disk-integrated observables. This work includes modeling EUV and X-ray spectral irradiance, predicting extreme events, and measuring flare energy distributions, using physics-informed AI/ML tools. We then apply these sun-as-a-star diagnostics to large photometric surveys, including the Roman Space Telescope Galactic Bulge Time Domain Survey.

Photometry, however, captures only part of a star's magnetic activity. It records flares but not the coronal mass ejections and winds that strike planetary atmospheres, and these eruptions have not been conclusively detected on any star other than the Sun. Their clearest signature is low-frequency radio emission (below about 30 MHz) from shocks and electron beams, which traces the outer corona and the stellar wind. However, the terrestrial ionosphere prevents ground-based observation of this emission. Lunar far-side observatories enabled by the Artemis program will provide the first sustained access to this last unexplored spectral window. We are developing the observational basis for ultra-long-wavelength (ULW) radio heliophysics, including the forward models and inversion methods needed to interpret these data and the effects of turbulent wind scattering on radio burst propagation.

We are also working on instrumentation, and our long-term aim is to build sentient telescopes: instruments that not only observe but understand what they see and decide what science to prioritize. On the hardware side, we are developing an integral field unit (IFU) spectrograph that records full EUV spectrum at every pixel in its field of view. It is designed to reach an unprecedented combination of spatial, temporal, and spectral resolution, delivering the spatially and time-resolved data on which the diagnostics above depend. On the software side, we are bringing artificial intelligence on board. Heliophysics foundation models such as Surya, which we helped develop with NASA and IBM, are being adapted to run on low-power edge hardware. Onboard, these models act as the instrument's brain: they anticipate transients before they occur, prioritize and compress data before transmission, and flag observations that warrant follow-up. By combining hardware and intelligence, we aim to enable observatories that can operate on the lunar far side, in deep space, and alongside future human missions, where the ability to think for themselves will be a necessity.