In the search for exoplanet biosignatures, researchers have obtained an unprecedented volume of atmospheric spectra in recent years; primarily due to the James Webb Space Telescope (JWST; online since 2022), with more data expected in the coming years and decades as several planned, ground-based, large telescopes come online. The relevant signal for quantifying a possible biosignature is the atmosphere’s spectrum: the amount of light transmitted through the atmosphere, at different wavelengths, when the planet passes in between its star and Earth. By determining which wavelengths are transmitted through the atmosphere, one can, in principle, determine which molecules are present in the atmosphere and then infer if those molecules imply the presence of life on the planet. There are challenges, however, to analyzing atmospheric spectra for the presence of molecules that signal life. First, current analysis models are too slow (computationally inefficient) and therefore only able to analyze a spectrum for the presence of one or two molecules at a time; this compared to a list of about 14,000 candidate biosignature molecules. Second, the library of known/tabulated molecular spectra is small, containing data for only a few hundred of the potential 14,000 biosignature molecules. Funds from this grant support a team led by Cecilia Garraffo, Director of the AstroAI Center at the Harvard/Smithsonian Center for Astrophysics to address both of these issues. Garraffo and her team will use advanced statistical techniques to iteratively improve a widely used analysis model, called POSEIDON, so that it can progressively analyze atmospheric spectra for many molecules at a time, rising eventually to an estimated 2000. In parallel, the team will use advanced machine learning techniques to develop, train, and validate an AI tool to predict how a molecule’s characteristics determine what sort of atmospheric spectra its presence would produce, adding an estimated 2000 molecules to the library that astronomers could use spectral analysis to search for. The effort, if successful, would lead to a significant improvement in our capacity to search atmospheric spectra for signs of extrasolar life.