Laser Powder Bed Fusion (LPBF) is the most widely used form of metal Additive Manufacturing (AM). Critical applications attempting to use metal AM include nuclear reactors, rockets, turbo-machinery and medical implants. Modern Design for AM enables high functionality, weight saving and even metamaterial properties. Furthermore no tooling is required, there is only a single production step and design may be changed based purely on software. However certification of structural integrity, essential for critical applications, presents a major problem. Traditional inspection methods such as X-ray Computed-Tomography are exorbitantly expensive, with inspection costs exceeding production costs. In-situ monitoring presents a solution by inspecting the part as it is built. While there are a wide variety of systems available none are proven to reliably detect defects in an industrial setting. This thesis aims to develop an in-situ monitoring system capable of economically detecting defects of concern to industry. The problem of defect detection was framed by determining the best metrics for measuring detection performance. Based on these metrics, the minimum viable performance for industry adoption was determined through consultation with industrial partners. To meet this standard, a system was designed to record coaxial melt pool images using a high speed streaming camera. This was chosen as it produces the largest volumes of data. While this made processing computationally intensive it maximised the volume of information available for detection. The outcome is a coaxial melt pool monitoring system which is capable of detecting (Probability of Detection (POD) = 0.8, Probability of False Alarm (PFA) = 0.1) small volumes of porosity (>0.2%) within localised regions (2mm voxels)...