Software bugs are ubiquitous and costly. An essential technique to reduce the cost of these bugs is static analysis, which allows software developers to reason about the run-time behaviour of their code without necessarily executing it. Unfortunately, in practice, software developers avoid using most static analysis tools, because they do not scale to modern software systems, report many false positives, and disrupt their workflow. In this talk, I will describe my journey for the past few years to address these limitations by improving three key aspects of static analysis tools: scalability, precision, and usability.