Data is “the new oil” because of its consequential transformations of virtually every industry in the first three decades of the 21st century. Data from the Internet of Things (IoT), which includes sensors and other smart electronics, is growing at an exponential scale. The number of IoT devices worldwide is projected to grow 7X from three billion in 2015 to 22 billion in 2025, and in a similar time, the amount of new data generated by IoT devices is expected to grow from 33 Zettabytes (ZB) in 2018 to 175ZB in 2025. However, existing networking infrastructure cannot keep pace with the growth of IoT devices and data, especially in rural areas of the United States. This dissertation examines how to build resilient rural networked systems to support agricultural IoT applications in this Zettabyte era and beyond. Our approach advances the state of the art by combining networked systems building and human-computer interaction (HCI). Specifically, we integrate system building, domain applications, and critical reflections on technical decisions in two complementary research phases. First, we create networked rural systems that are vendor agnostic, resiliently allowing IoT data transmission, collection, and analysis, while autonomously detecting and repairing failures. We present the Software-Defined Farm (SDF), a cloud-based architecture that advances the state of the art towards extensible, reconfigurable, and fault-tolerant IoT systems. Critically, we upscale our technical systems investigation with HCI by using the SDF implementation and deployments to analyze how researchers envision and orient toward the eventual use of such systems in rural areas. Our analysis shows a consequential contrast between researchers' seamless visions of rural networked system deployments and the seamful realities of their own system building. Second, building on insights from the first part of this dissertation, we investigate how to scale resilient networked systems across multiple geographically separated sites. We design, implement, deploy, and evaluate Ursula, a general-purpose system that centrally manages applications deployed across independent, intermittently available farm IoT devices. We use our technical deployments to analyze how institutional and disciplinary factors influence researchers' envisioned societal impacts. Our analysis demonstrates how, even when computing researchers envision particular impacts, they do so within organizational practices that gradually reshape not only the envisioned impacts, but also the work towards them. In concluding each dissertation part, we show that our integrated approach of technical and sociotechnical aspects is one of the core contributions from this dissertation. Specifically, we demonstrate that the networking and HCI advancements presented together in this dissertation would not be possible without the integrated approach. We call the approach trilingualism, which entails the integration of deep technical work, its domain applications, and critical reflections on the technical decisions. Together, trilingualism provides a step in advancing the state of the art in networking for the Zettabyte era.