Organic compounds containing carbon, hydrogen, and oxygen (CHO) are not only foundational to numerous industrial applications but also fundamental to life and chemistry across the universe. Their versatility arises from the unique bonding properties of carbon. However, developing a single predictive theoretical framework to describe these chemically diverse systems remains a significant challenge for atomistic modeling, especially due to limited experimental access to interatomic interactions. Circumstellar environments contain a diverse range of organic compounds, ranging from small molecules to complex solid-state structures. These include fullerenes, carbon chains, hydrogenated amorphous carbon, and polycyclic aromatic hydrocarbons (PAH) [1]. The large-scale processes of the formation of stars and planets, and the evolution of galaxies, involve fundamental physical and chemical processes that drive the transformation of organic matter. Understanding their origins and formation mechanisms remains a key open question. In this talk, I will discuss how machine learning interatomic potentials (MLPs) provide a unified framework to describe organic materials within a single model, bridging applied materials science and isolated molecules under extreme astrophysical conditions. First, I will present our general-purpose CH MLP [2], developed using the Gaussian Approximation Potential (GAP), and demonstrate its application to the formation of simple and complex alkanes, aromatic hydrocarbons, hydrogenated amorphous carbon (a-C), and CH systems under extreme conditions. I will show how this model generates hydrocarbons of varying size and complexity without relying on prior knowledge of organic chemistry, thereby capturing the relationships and transformations among hydrocarbons and hydrogenated carbon materials across multiple conditions within a unified computational framework. Next, I will address efficient strategies for sampling training data to extend this framework to a broader range of organic materials composed of C, H, and O, including organic molecules, cellulose, and its derivatives [3]. Finally, I will show how we incorporate experimental observables into simulations and tailor the MLPs to aid interpreting astrophysical observations. [1] S. Kwok, A. Astrophys Rev, 24, 2016. [2] R. Ibragimova, M.S. Kuklin, T. Zarrouk, M.A. Caro, Chemistry of Materials, 37(3), 2025. [3] T. Zarrouk, R. Ibragimova, A.P. Bartók, M.A. Caro. JACS, 146, 2024.