The microlensing mission planned on the Roman Space Telescope, scheduled to launch in the mid-2020s, is expected to revolutionize our understanding of exoplanet demographics. Owing to the pathological parameter space of binary microlensing which contains a multitude of local likelihood maximas that are narrow and deep, current analysis of such events are done on a case-by-case basis with computationally expensive "grid-searches" required as a prerequisite to MCMC posterior sampling. While the status-quo approach could sustain on the order of the dozens of events discovered each year, it creates a significant challenge for the Roman microlensing survey which is expected to discover thousands of planetary microlensing events. In this talk, I will present a new likelihood-free inference approach named amortized neural posterior estimation, where a neural density estimator (NDE) learns a surrogate posterior as an observation-parameterized conditional probability distribution, from pre-computed simulations over the full prior space. After training on mere ~300,000 Roman-like 2L1S light-curve simulations, the NDE automatically produces accurate and precise posteriors for any future Roman light-curve within seconds, thus allowing for fast and automated inference. The NDE also captures expected posterior degeneracies. The NDE posterior could then be refined with downstream MCMC to produce the exact posterior solution with minimal burn-in steps, reducing the computation expense by orders of magnitude.