Discovery and engineering/optimization of proteins with novel functions is a laborious process relying on the design-build-test cycle to reach desired levels of protein fitness. Whether the function is binding to a specified target moiety or improved catalytic function, in vitro methods for discovery and directed evolution require the generation of DNA variant libraries. Legacy approaches for generating DNA variant libraries are often associated with shortcomings related to composition bias, or inclusion of theoretically non-interesting members containing one or more undesirable sequence features (liabilities). The silicon-based, high-throughput DNA synthesis platform at Twist Bioscience can be leveraged for generating highly complex, targeted variant libraries with unprecedented control, uniformity, and quality. This technology enables researchers to increase effective library sizes, to include the diversity most interesting to the project or protein of interest, and to remove sequence liabilities and custom motifs from the library design. Here we highlight Twist Bioscience’s ability to construct complex libraries of high-diversity confined to defined regions within a gene (for example, CDRs), or with diversity scattered along the length of a gene. These NGS-verified libraries provide a powerful solution for precise and efficient interrogation of the variant space, and enable users to make iterative designs for discovery and optimization.