Single-walled carbon nanotubes (SWCNTs) have garnered significant interest for their applications in optical biosensing. SWCNTs exhibit a sensitive and photostable fluorescence in the near-infrared (NIR) range, allowing for the continuous monitoring of target analytes in vivo with minimal absorption from biological tissue. However, despite their advantages, the behavior of these nanomaterials can be difficult to predict, especially for bioconjugates, such as DNA-wrapped SWCNTs (DNA-SWCNTs), whose interactions are often insufficiently characterized and poorly understood. When non-covalently functionalized with single-stranded DNA, SWCNTs can exhibit improved colloidal stability, biocompatibility and sensing capabilities towards a wide collection of target analytes. These DNA-SWCNT complexes are highly versatile as their optical properties can be modulated by changing the DNA sequence. However, the relationship between the DNA sequence and the optical properties of the resulting DNA-SWCNT complex is unknown, hindering the ability to engineer these complexes in a predictive manner. As a result, the design of DNA-SWCNT sensors currently relies on empirical approaches that often result in sensors with compromised performances. In this thesis, we take inspiration from strategies developed in the field of protein engineering to engineer the DNA-SWCNTs in a more controlled and directed manner. Using directed evolution, an approach extensively applied for engineering proteins, we engineer improved DNA-SWCNT complexes through iterative cycles of mutation, screening and selection. We first apply this strategy to significantly improve the fluorescence intensity of an existing DNA-SWCNT sensor while retaining the sensor performance. We then extend this approach to engineer new sensors by tuning different sensing properties, such as selectivity and sensitivity, in a chirality-dependent manner for the multimodal detection of mycotoxins in food products. Building on the knowledge acquired from the mutants generated during these previous iteration cycles, we establish DNA design rules allowing us to engineer a new generation of DNA-SWCNT sensors through rational design. We additionally expand our study to explore alternative approaches to further modulate the properties of DNA-SWCNTs beyond the capabilities of natural DNA oligonucleotides through the incorporation of artificial chemical modifications. Finally, we conclude with scalable and high-throughput approaches for mutating, amplifying, purifying, and conjugating DNA sequences to SWCNTs to increase the sizes of the libraries that can be screened for directed evolution. The approaches presented in this thesis therefore provide a means of tackling bottleneck challenges in the design of DNA-SWCNT complexes through the exploitation of bioengineering strategies.