A major challenge of exoplanetology is to combine the observed diversity of planetary systems in a consistent theory of planets formation and evolution. In order to guide us in this objective, both new discoveries and detailed characterization of peculiar planetary systems bring crucial constraints. This thesis develops these two aspects. On one side, I present my implications in the discovery of new planets with the CORALIE, HARPS-N and SOPHIE high-resolution spectrographs. On the other side, I modeled the dynamics of several multi-planet systems, and developed a technique that uses orbital stability concepts to refine the planet parameters and improve our knowledge of the multi-planet systems. This double process is key to select wisely the most promising systems for in-depth observations with the most powerful telescopes to come, getting us closer to this consistent theory.