Recent advances in manufacturing and growing concerns on the sustainability of aviation environment have led to a remarkable interest in electrical unmanned aerial systems (UASs) in the past decade. Among various UAS types, the newly designed quadrotor biplane tailsitter class is capable of delivering a wide range of civilian and military tasks, relying on its Vertical Take-Off and Landing (VTOL) capability as well as great maneuverability. Nevertheless, as such UASs employ rotors to generate thrust, and wings to generate lift, and operate at less-understood low to mid-Reynolds flow regime, they experience complicated flight aerodynamics with a noise generation mechanism which is different from common aircrafts. The present work aims at addressing this knowledge gap by studying the aerodynamics and aeroacoustics of a UAS of this type designed by the Army Research Lab. High-fidelity computational fluid dynamics (CFD) simulations are carried out for a wide range of operating conditions to understand the physics involved in the UAS aerodynamics and characterize its performance. Relying on the CFD results, a physics-informed reduced order model (ROM) is developed based on machine learning algorithms, to predict the propellers effects on the wings and calculate the dominant loads. The results of this study indicate that the UAS aerodynamics is significantly influenced by the propeller-wing interaction, which makes it challenging to estimate the loads by classic methods. The proposed physics-informed ROM shows a promising performance based on its computational cost and accuracy. Additionally, it is found that the aeroacoustics of the UAS is ruled by a two-way mechanism through which the propellers and the structure impose unsteadiness on each other.