Accurate measurement of exoplanetary masses is a critical step in addressing key aspects of NASA's science vision. Measuring masses of Earth-analogs around FGK stars out to 10 pc requires sub-microarcsecond astrometric accuracy, which is not within the capabilities of current instrumentation. Thus, new technology will be required to build an astrometric instrument capable of achieving such performance. This will immediately empower the possibility for dedicated astrometric missions, and perhaps most enticing, it will enable astrometric observing modes to be added (with relatively low cost and impact) to any mission boasting a sufficiently stable direct imaging platform. In this paper, we provide an overview of the scientific goals and technology utilized on two of NASA's astrometry testbeds dedicated to advancing stellar astrometry for exoplanet detection. The first one, located at the Jet Propulsion Laboratory (JPL), is dedicated to imaging stellar astrometry on sparse fields. The goal of this testbed is to mature the Diffractive Pupil (DP) technology to TRL-5, demonstrating high-fidelity performance in a relevant environment. This testbed operates in a vacuum tank at the High Contrast Imaging Testbed (HCIT) at JPL. The second testbed, located at NASA Ames Research Center, is dedicated to advancing narrow angle relative astrometry to detect exoplanets around nearby binary stars. The key technology in this testbed is a DP specially designed to measure the angle between two sources on the sky. This testbed operates in air and aims to bring this technology to TRL-4.