Without knowing the subsurface condition, engineers cannot design because for a single infrastructure and a building, the most important part is the foundation. But conventional practice for collecting data is really expensive We want to investigate the possibility of using machine learning and mixed reality technologies like AR and VR to help improve the process of geotechnical subsurface exploration. We try to make the whole process become automatic. People can reuse some exploration data, which is archived in their warehouse. So we can digitize all the plans, digitize all the blueprints, then we can synthesize all this data together by using motion-learning techniques and also the computer-aided design techniques. So we can reconstruct the 3D condition of a subsurface. Also, we can use stochastic simulation techniques to reduce uncertainty. Then we can use this model to do the analysis, which make the process very fast and very accurate. so that we can provide engineers with a precise number of their design confidence so they can confidently design their structures. We can use VR and AR technology to make this whole process become more dynamic. Things become interactive. The experts, they can wear the VR headset to navigate in the environment and they can interactively choose the spot of the subsurface that they have interest in. They can zoom in and see the 3D data immediately. This can dramatically improve the whole process with much higher efficiency and effectiveness. This is a future trend. Not only in subsurface exploration, but in any field. People will interact with a system more and more seamlessly, more precisely, and more conveniently.