The introduction and widespread adoption of Artificial Intelligence and data-driven applications, i.e. software systems that rely on large sets of data to define their functionality, have created a paradigmatic shift in many areas of our lives, not just for end users but particularly for professional software developers. In this dissertation, we take a human-computer-interaction perspective on how technologies like Machine Learning have affected software developers, their work practices, tools, methods, etc. To this end, we consider two perspectives: First, just as data-driven applications have enabled new types of applications and solutions to challenges in many areas, they have also given software developers new tools and approaches for writing and creating software. For example, natural language processing has yielded systems for quickly generating working code from informal descriptions. Besides new tools, applying Machine Learning techniques to the wide landscape of existing development tools also promises to improve the work of developers, e.g. with personalized and adaptive tools. Thus, the first half of this dissertation will be dedicated to investigating how data-driven applications can be applied to enhance the software development experience. However, for developers, the widespread adoption of data-driven applications has a second consequence: the kind of software they write changes to incorporate these data-driven functionalities. Since there are paradigmatic differences in how data-driven systems operate, compared to more traditional software systems, this naturally also affects how they are developed and what their underlying code looks like. Therefore, the second part of this dissertation is first concerned with these differences from a human-centered perspective, particularly how developer behavior differs when working on the development of data-driven and traditional software systems, focusing on code reading. Since this demonstrated that data-driven development has its unique challenges, we then consider how tool support can help developers in these changing circumstances. To this end, we explore the literature on tooling. This revealed both evolving and new tools for this new reality but also highlighted areas where development tools can still be improved and tailored to the requirements of data-driven development. Thus, we finally explore new tool implementations to address the challenges of data-driven development as well as the needs and wishes of developers that we elicited throughout this work. Insights from these two parts demonstrate how the changes in the underlying technology also drive changes in the human perspective. As the technology continues to evolve at a growing pace, it will become equally important to continuously assess this relationship to address any potential mismatch, ensure the continued quality of software in the face of ever-increasing complexity, and ensure developers can continue to effectively leverage these technologies to shape our lives with technology. This work therefore can only contribute only a momentary assessment of the state of software development and a foundation for future research into how data-driven applications continue to affect software developers.