Identifying electrons and muons in the complex environment of LHC collisions is crucial for precision measurements and the discovery of rare processes. For instance, the recent observation of the simultaneous production of four top quarks – one of the rarest and heaviest processes in the standard model – was significantly aided by a special strategy for lepton identification. In the first part of this talk, I will present a machine learning-based algorithm developed to enhance the identification performance of light leptons. As we gear up for the high-luminosity phase of the LHC, projected to deliver a ten-fold increase in data by 2040 compared to the existing dataset, the CMS experiment will have major upgrades across several detector components. I will discuss the electrical characterization of silicon sensors designed for the upgraded tracker and endcap calorimeter. I will also briefly outline my current engagement in the thermal qualification of the mechanical structure, with an integrated cooling system, that supports detector units in the outer tracker endcaps. I will conclude with some novel ideas for detector development aimed at future collider experiments.