The nature of dark matter (DM)—constituting approximately 26.8% of the universe's mass-energy density—remains one of the most fundamental unsolved problems in cosmology. Dark matter is a non-luminous form of matter whose existence has been inferred through its gravitational effects. Initially proposed as massive particles interacting weakly, its persistent evasion of direct detection over decades has intensified interest in alternative models. Particularly compelling are models of light DM (sub-GeV), hypothesized to interact with electrons via hidden-sector mediators. I present a detailed account of my contributions to the search for light DM using silicon charge-coupled devices (CCDs) within the DAMIC-M (Dark Matter in CCDs-Modane) experiment. I first outline theoretical motivations and observational evidence supporting DM’s existence. Subsequently, I describe the detection principles and operational mechanisms of DAMIC-M’s CCD detectors. I highlight the operational advantages of Skipper CCDs, which perform repeated non-destructive charge measurements (NDCMs) to achieve an unprecedented reduction in readout noise. This capability enables us to count single electrons, lowering the energy threshold crucial for sensitivity to previously inaccessible light DM interactions. A key aspect of this work is the rigorous CCD characterization program developed for DAMIC-M. This program implements stringent quality assurance methodologies at every stage of CCD selection, ensuring optimal detector performance. Such careful selection is critical for achieving DAMIC-M’s ambitious data acquisition goal of acquiring one kg-year exposure with a low single-electron rate. The Low Background Chamber (LBC), a prototype for DAMIC-M, is a focal point of this dissertation. I present detailed descriptions of the LBC’s detector design, environmental controls, radioactive background mitigation strategies, and key results from background measurements. The LBC enabled high-precision data taking and facilitated direct detection searches to establish world-leading constraints on dark matter-electron interactions within the 0.5–1000 MeV/c^2 mass range at that time. Furthermore, meticulous background characterization is emphasized, employing spatial coincidence analyses and event classification methods to identify and mitigate radioactive contamination sources, paving the way for DAMIC-M to achieve its <= one event/eV/kg/day background goal. This work also evaluates advanced CCD processing methods to reduce 210Pb surface contamination relative to earlier DAMIC detectors at SNOLAB. Lastly, I introduce the "Pocket Pumping" technique to characterize and mitigate charge traps within CCDs. In summary, this work encapsulates my research program--from theoretical motivations and detector development to intricate analyses and crucial experimental results—demonstrating DAMIC-M’s capability to explore previously inaccessible regions of dark matter parameter space.