Searches for invisible decays of the Higgs boson, produced via vector boson fusion, are presented in this thesis.
Data from proton-proton collisions at \sqrts, collected with the CMS detector during 2016 and corresponding to an integrated luminosity of ${\SI{35.9}{fb^{-1}}}$, are used.
The data are in agreement with the standard model prediction, and the observed (expected) upper limits on the invisibly decaying Higgs boson branching ratio \brhinv are $0.58(0.30)$ and $0.33(0.25)$ at $95\%$ CL for the `cut-based' and `shape-based' approaches, respectively, as published in \ReferenceRef{Sirunyan:2018owy}.
Several combinations with other relevant analyses are performed using data from Run-1, Run-2 (2015), and 2016 to further improve the sensitivity to \brhinv, and the results are published in Refs.\xspace\cite{Sirunyan:2018owy,Khachatryan:2016whc}.
The observed (expected) upper limit on \brhinv is set at $0.19(0.15)$ at $95\%$ CL from the combination performed with all datasets, and it is the most stringent limit to date.
Interpretations of the result under non-SM production assumptions and Higgs-portal models are also provided.
The observed (expected) $95\%$ CL upper limit on $\frac{\sigma}{\sigma_\text{SM}}\times\brhinv$ varies in the range $[0.14,0.24]([0.11,0.19])$, assuming non-SM production cross-sections.
The observed $90\%$ CL upper limit of $\brhinv < 0.16$ is translated into an upper limit on the spin-independent DM-nucleon elastic scattering cross-section.
This limit is the most stringent constraint for $m_{\chi}<{\SI{18}{GeV}}$ or $<{\SI{7}{GeV}}$ assuming a fermion or a scalar DM candidate, respectively.
\newline
R\&D computing activities within the Computing \& Offline Project at CMS are presented in this thesis.
The results from the pioneering implementation of the dynamic on demand analysis service (DODAS) of CMS in different cloud environments and the benchmarking of high energy physics use-cases are an important milestone in CMS, as published in \ReferenceRef{Spiga:2650034}.
The feasibility studies for the integration of production workflows are some examples to which the work in this thesis significantly contributes.
Moreover, the exploitation of cloud resources from UK's Amazon Web Services demonstrates the full exportability of DODAS, which thus provisions on-demand resources in the Grid without any constraint on their physical location.