The work performed during the whole period of the project has been first a work of acquisition of skills related to the theoretical study of the dynamics of nonlinear fiber cavities. After this training, various numerical simulations have been performed to engineer a fiber nonlinear cavity capable to sustain temporal solitons characterized by specific features, and in particular by a short pulse duration, compared to what already present in literature. Experimental skills have been acquired related to the process of stabilization of fiber cavities and of soliton addressing in those cavities. Then it has been studied the possibility of experimentally implementing nonlinear fiber cavities in in-line configuration. Some technological issues have been addressed, specifically the realization of mirrors suitable for the implementation of those in-line cavities. A preliminary study has been performed with the scope of fabricating dielectric stacks on the facets of a specific type of optical fiber, to be used as reflectors for the implementation of the compact all-fiber resonator. Alternative solutions have been also found to circumvent potential practical difficulties in the realization of those dielectric stacks, such as for example the adoption of fiber Bragg gratings as input and output couplers/reflectors for those cavities. A setup for thermal poling of optical fibers and a setup for UV erasure of second order nonlinearity have been also implemented with the scope of fabrication of new samples of periodically poled fibers. The most important scientific results of the project are summarized in a publication in Nature Photonics, where it has been reported the first all-fiber optical parametric oscillator (OPO) for the generation of parametrically driven Kerr cavity solitons. In this paper the periodically poled fiber, which represents the base of the technological platform proposed in my project, has been adopted in conjunction with an active fiber (Erbium doped fiber) to generate solitary waves suitable for the application such as random bit generation and Ising machines. The results of this work have been disseminated also in international conferences and workshops. The relevance of this work stands in the fact that all the results obtained will be re-used for a future implementation of a compact all-fiber resonator, which will need to be modified to include optical gain.