Optimization of existing domain or traditional antibodies is aimed at improving desired biological properties such as binding affinity, pH-dependent binding preference, expressibility, or developability. We use several complementary techniques in combination to introduce targeted diversity into the antibody sequence. Our techniques include in vivo antibody evolution via inducible hypermutation in CDR regions and combinatorial antibody libraries. We have perfected a robust platform to optimize existing antibodies to targeting the tumor micro-environment and reducing the potential on-target off-tumor side effects via developing antibodies that have high affinity at low pH (pH in the tumor microenvironment) but have low or weak target binding at the normal pH. Such pH dependent antibodies would be desirable for treatment of solid tumor, especially in T-cell mediated immunotherapy. For optimization of human IgG antibody, we use both heavy chain or light chain shuffling using pre-made human heavy chain or light chain libraries, respectively, and combinatorial libraries with defined amino acid substitutions at selected sites. Yeast clones expressing antibodies with desired attributes are isolated by FACS followed by individual clone validation.