Single-walled carbon nanotubes (SWCNTs) generate near-infrared fluorescence that is useful for biomedical and environmental optical sensing. They can be colloidally suspended by π-electron-rich biopolymers like DNA that self-assemble onto their surfaces. The optical properties of these DNA-wrapped SWCNT (DNA-SWCNT) sensors, such as the sensitivity of the fluorescence to specific analytes, vary with the sequence of the DNA wrapping. A few studies have studied the relationship between the DNA sequence and its effects on SWCNT fluorescence. However, the vast DNA sequence space is unexplored, and the unpredictable effects of the DNA sequence on the sensor properties remain a bottleneck for engineering these sensors. This chapter focuses on high-throughput techniques inspired by directed evolution to overcome this bottleneck. We review the application of these techniques to engineer different optical properties, such as brightness, selectivity, and sensitivity, with little to no information on the DNA-SWCNT/analyte interactions. This chapter further examines complementary computational approaches using machine learning-based pattern recognition for the semi-rational design of these sensors. The combination of these computational tools with high-throughput strategies provides a powerful basis for not only engineering DNA-SWCNT sensors but also understanding their function.