Establishing scalable nanomaterial synthesis protocols remains a bottleneck to commercialization and is therefore the subject of intensive research and development. Here we present an automated, machine-learning microfluidic platform capable of synthesizing optically active nanomaterials from target spectra theorized or derived from published prior experience. Implementing unsupervised Bayesian optimization as a Gaussian process reduces optimization time and reduces the need for prior knowledge to start the process. PTFE tubing and connectors allow for easy reactor design changes. Ultimately, this platform replaces labor-intensive trial-and-error synthesis and provides a path to standardized and high-volume synthesis, slowing the translation and commercialization of high-quality nanomaterials. As a proof of concept, Ag nanoplate and Prussian blue nanoparticle protocols were optimized and validated for mass production.