The SPEED project addresses the problem of computing symbolic computational complexity bounds of procedures in terms of their inputs. We discuss some of the challenges that arise and present various orthogonal/complementary techniques recently developed in the SPEED project for addressing these challenges.