Radio interferometric imaging traditionally treats calibration and imaging as separate steps, using packages, such as AIPS, CASA, and DIFMAP. However, both processes rely on shared prior knowledge about the source and instrument. In this talk, I present a probabilistic calibration and imaging approach based on Bayesian imaging software resolve which jointly infers calibration and imaging parameters, naturally propagating calibration uncertainties into the final image reconstruction. Using variational inference to mitigate computational costs, this method achieves high-fidelity and super-resolved images while providing uncertainty estimates for calibration and imaging results. The framework is extended to include polarimetric calibration, enabling robust full-Stokes imaging. I demonstrate the method using VLBI observations of active galactic nuclei (AGNs), including: M87 (VLBA and GMVA+ALMA, 43 and 86 GHz), revealing the black hole shadow and jet with improved resolution and quantified uncertainties in ring parameters, 3C273 (VLBA, 15 GHz), OJ287 (GMVA+ALMA, 86 GHz), and 3C120 (VLBA, 22 GHz) , producing reliable polarimetric images, instrumental calibration solutions (gains, D-terms), and rotation measure. Additional applications include the radio galaxy Hydra A (VLA, 2.5 GHz) and the microquasar SS433 (VLA, 5 GHz). I also present a Bayesian multi-frequency imaging method with frequency phase transfer (FPT) observations 4C39.25 (KaVA+Yebes, 22 and 44 GHz), enabling improved UV coverage and core shift measurements. These results highlight the potential of Bayesian calibration and imaging to provide reliable reconstructions with quantified uncertainties, offering improved observational constraints on the innermost structures of black holes and their jet formation, and promising applications for next-generation radio interferometers.