<p dir="ltr">This thesis presents two studies in search for violation of lepton flavor universality as predicted by the Standard Model. The first searches for signs of the violation by studying beyond the Standard Model (BSM) physics models involving same flavor and opposite sign dilepton pair and bottom quarks as final states. This study was done using the dataset collected during years of 2016, 2017 and 2018, with center of mass energy $\sqrt{s} = 13$ TeV and integrated luminosity of 138 $fb^-1$. In the study, scale factors were derived in order to correct deviations between simulation and real life data, specifically for high transverse momentum muons and top\&anti-top quark background. Furthermore, lower limits of energy scale were calculated leading to exclusion of the BSM models with energy scale values lower than that of the calculated value.</p><p dir="ltr">The second study also searches for of the lepton flavor universality violation, but in the specific decay of a tauon into three muons ($\tau \rightarrow 3\mu$). In the study, graph based neural network model (GNN) designed to classify $\tau \rightarrow 3\mu$ events at the CMS detector was converted to high level synthesis (HLS) code, so that the GNN could be coded into a custom hardware such as field programmable gate arrays (FPGA) for deployment. Moreover, techniques such as pruning and quantization were applied in an attempt to make the GNN more light weight, due to strict requirements of FPGA.</p>