Easy processability, tunable mechanical and optoelectronic properties, and the endless possibilities of molecular modifications enable conjugated polymers (CPs) to be used in a wide range of lightweight, low-cost, and flexible organic electronic devices. However, despite the tremendous efforts in molecular engineering and improved electronic properties, the thermomechanical-structure-property relationship of these semiconducting materials is still less investigated. Predicting the thermal, mechanical, and photovoltaics (PV) properties of CPs is challenging due to heterogeneous chain architectures and diverse chemical building blocks. To address this critical issue, this dissertation aims to employ novel multiscale modeling and data-driven approaches to characterize the thermomechanical and optoelectronic properties of CPs. In particular, we utilize machine learning (ML) and scale-bridging molecular modeling techniques scaled from quantum mechanics simulations to force field all atomistic molecular dynamics (AA-MD) and coarse-grained (CG)-MD simulations to explore the multifunctional behavior of CPs. Validated by experimental measurements, our AA-MD and CG-MD simulations can capture glass transition temperature (Tg), elastic modulus, and strain-induced deformation mechanism of the CPs with varied side-chain lengths and backbone moieties. Furthermore, through the integration of ML, AA-MD simulations, and experiments, we propose a simplified but accurate predictive framework to quantify Tg directly from the geometry of the CPs’ repeat unit. Next, first-principle calculations developed upon density functional theory (DFT) are utilized to quantify electronic configuration changes of CPs due to the charge injection. Finally, created upon ab initio excited state dynamics, we report a computational methodology to explore the PV performance of donor-acceptor organic bulk heterojunctions (BHJ). This computational framework facilitates screening of the best donor-acceptor molecules and narrows down the list of potential candidates to be used in BHJ. We believe data-driven and multiscale modeling approaches established in this dissertation are important milestones for the design and structure-property prediction of CPs and organic BHJ, paving the way for developing the next generation of organic electronics.