Second harmonic generation microscopy is used to image interfaces or non-centrosymmetric structures and molecules. Two-photon fluorescence microscopy can image molecules within cells without using markers. Thus those technics have lead to advances in our understanding of structure such as lipid membranes, or of the physical properties of tightly confined of water. They are also powerful tools for biological imaging, allowing to study microtubule orientation or electric potential in neurons for instance. However Multiphoton emission microscopy, as it relies on non-linear light-matter interaction, requires high intensity. Femtosecond lasers allow for such intensities while keeping fluence low enough to not cause damage to the material. Making such a laser tunable though the use of optical parametrical amplification (OPA) improves penetration in tissues and water, and allows to probe some materials at resonance. Fiber lasers should be well suited for this application, as they are cheaper than bulk lasers and less sensitive to misalignment. During this PhD we aim at building a new tunable femtosecond fiber laser. We will seek to optimise the parametrical amplification process to have high efficiency conversion from 2500 to 700nm. We expect to reach pulse duration as short as 30 fs, energy up to 200nJ and repetition rate up to 2MHz.