Traditional Searches for Extraterrestrial Intelligence (SETI) use large radio telescopes to look for artificial signals from specific stars. In the era of large astronomical surveys, it is now possible to efficiently search for objects having certain predicted signatures of astro-engineering. In this talk, I present an international, cross-disciplinary project between astronomers and researchers in machine learning, the "Vanishing and Appearing Sources during a Century of Observations" (VASCO) project, where we scan the sky for objects that have physically disappeared from (or appeared on) the sky during the last decades. Some of the contaminants we expect are variable astrophysical objects with decade-long time scales. We compare the USNO to PanStarrs catalogues using several epochs of observations. Here, I present the updates about the candidate from the pilot paper (Villarroel+ 2016) and the discovery of ~100 transients (~1/3rd with amplitudes larger than 5 mags). The final goal of the project is to identify interesting astrophysical targets for follow-up analysis with extreme, exotic or bizarre patterns of variability.