This dissertation explores socio-spatial interfaces—interactive systems embedded in everyday environments that compute social parameters such as proxemics, orientation, and gaze, and translate them into spatial and technological cues to facilitate social connection. In an era shaped by digital saturation and social fragmentation, it argues that technology should help reconnect us with our innate humanity for genuine social connections: supporting embodied and situated interaction, and fostering feelings of togetherness in public life. Drawing from architecture, human-computer interaction (HCI), social psychology, and artificial intelligence (AI), this work develops a conceptual grounding for socio-spatial interfaces, and demonstrates its application in the context of stranger interaction through two design exemplars. The first, SocialStools, is an extended reality installation embedded in stools that uses audiovisual feedback to promote proxemics-based interaction. Building on behavioral insights from a lab study, the dissertation introduces a behavioral framework for analyzing ice-breaking dynamics and applies it in a field deployment. It further explores how such systems can transition from interactive to adaptive using reinforcement learning. The second exemplar, MirrorBot, explores how a mobile robot with repositionable mirrors can redirect visual relationships in space to spark serendipitous mutual awareness, further articulating the design space of socio-spatial interfaces. Together, this dissertation repositions space not as a passive backdrop, but as an expressive, co-constitutive participant in human-computer interaction. It offers a new direction for human-computer interaction, providing theoretical grounding and design knowledge for technologies that nurture connection, curiosity, and community in the shared spaces of everyday life.