This manuscript presents highlights of research I contributed to on magnetic random access memory (MRAM) since arriving at the CEA-Grenoble Spintec laboratory in 2002. This work involved the development of MRAM concepts using magnetic field and spin transfer torque (STT) writing methods and the materials necessary to demonstrate their principle of operation. After briefly summarizing my career path in the first chapter, I introduce areas that would benefit from the adoption of non-volatile memory in particular as embedded memory. The second chapter reviews the advantages of using the current flow through the cell in thermally assisted writing using magnetic fields and STT. It focuses on the heating dynamics and methods to determine the temperature increase inside a magnetic tunnel junction pillar. The next chapter, reviews solutions to write and reverse the magnetization direction at the ultimate speed limit defined by the magnetization precession frequency in the GHz range. It also discusses the tunnel junction barrier breakdown at short pulse widths. This is followed by a close look at the advantages of perpendicular magnetic anisotropy materials for MRAM, as the concept most likely to gain industry-wide adoption, and the material developments necessary to scale cells below 20nm diameter. Finally, I summarize our results in the context of possible future research directions in the field of MRAM.