This document describes a Bayesian inversion approach to jointly interpret multiple seismic data types that provide information about anisotropic layering in the upper mantle. Surface wave dispersion curves, SKS splitting measurements, and receiver functions are traditionally interpreted separately, but sample different volumes of the Earth and have different sensitivities and uncertainties. The proposed method directly inverts seismograms for SKS and P phases using a cross-convolution approach, avoiding intermediate processing steps. A transdimensional Markov chain Monte Carlo scheme obtains probabilistic 1-D seismic velocity profiles down to 350 km depth beneath two stations, treating the number of layers and presence of anisotropy as unknown parameters. For both stations, the lithosphere-asthenosphere boundary is clearly visible and marked by