I think you underestimate the destructive impact of 30 failed interviews. Stepping back and just doing algorithms for some time might be a good thing to do at this point.