The aim of the work presented herein was the generation of high-affinity shark vNAR domains targeting different disease related antigens. For this, the natural IgNAR V domain repertoire of the bamboo shark (Chiloscyllium plagiosum) was analyzed and in analogy to these findings, a semi-synthetic complementarity determining region 3 (CDR3)-randomized Type IV vNAR library was constructed for yeast surface display. Through library screening against several disease-related antigens multiple different antigen-binding vNAR domains were isolated and characterized in terms of affinity, revealing moderate affinities in the triple-digit nano-molar to micro-molar range. For optimization of antigen-binding vNAR domains, a new methodology for the affinity maturation was established that relies on the diversification of CDR1 of target-enriched binders. Sub-libraries were constructed in which five residues of the CDR1 loop were randomized and affinity-enhanced vNAR domains were identified by library screening using significantly decreased target concentrations. Affinities determined using yeast surface display revealed substantially affinity-optimized clones compared to parental molecules, obtained from initial library sorting. Additionally, several vNAR domains were produced as soluble protein and characterized more meticulously in terms of affinity using bio-layer interferometry and in terms of stability via thermal shift assays. In this respect, affinities calculated by yeast surface display strongly correlated with affinities determined for soluble IgNAR V domains. Moreover, all produced vNAR variants exhibited high thermo-stability. Besides, an EpCAM-binding, affinity-matured vNAR domain was expressed as Fc-fusion protein in mammalian cells. Characterization of this formatted variant using bio-layer interferometry resulted in a moderately, but significantly enhanced affinity by the factor of three, presumably through avidity-effects. Furthermore, also the generation of a new antigen-binding site into the IgNAR variable domain was in the scope of this work. This was achieved through the diversification of hypervariable loop 2 (HV2) of the IgNAR V domain and library screening. An EpCAM-specific vNAR was used as starting material and nine residues in HV2 were randomized. Target-specific clones comprising a new HV2-mediated paratope were isolated against cluster of differentiation 3ε (CD3ε) and human Fcγ while retaining high affinity for EpCAM, resulting in bi-specific vNAR molecules. Essentially, it was demonstrated that a new paratope can be engineered into the vNAR scaffold that acts independently from the original antigen-binding site, composed of CDR3 and CDR1, as verified by yeast surface display.