There has been limited research in computational aspect of change-point detection, to achieve exact error control, especially in the high-dimensional setting. We address the computational challenge of change-point detection using the recently developed convex optimization based hypothesis testing techniques developed in A. Juditsky, A. Nemirovski, "Hypothesis testing via affine detectors," Electronic Journal of Statistics 10:2204--2242, 2016. Given an evolving in time noisy observations, we would like to decide, in a sequential fashion, whether there has been a change. We develop "computation-friendly" sequential decision rules and demonstrate that in our context these rules are provably near-optimal. This is a joint work with Vincent Guigues, Anatoli Juditsky, Arkadi Nemirovski, and Yang Cao.