During my presentation, I will showcase how my research group brings together the two fields of deep learning and neuroscience in mutually beneficial ways. In the first part of the talk, I will demonstrate how the development of bio-plausible deep learning models not only aids neuroscientists in formulating novel hypotheses about the brain but also provides new avenues to tackle significant challenges in deep learning. In the second part of the talk, I will present our experimental neuroscience findings on how sensory information is transformed into goal-directed behaviors in the mammalian brain. I will then describe how such neuroscience discoveries can be used to develop novel AI technologies with a focus on action transformer networks that control (virtual) robotic agents.