Most programmers don't start projects from scratch, but inherit them. Often, we join large, messy, years-old code bases that are poorly tested and documented, and are intimidating to work with. Working with them can feel like tap dancing through a mine field. Generative AI also provides challenges for software developers, as it can generate huge amounts of code that need to be understood. This talk offers practical strategies for navigating and improving legacy code. We'll see techniques like using git history with scripts to identify software hot spots, introducing software "seams" for safe refactoring, and isolating risky changes through "sprouting" and "wrapping" techniques. You'll see how small, strategic changes can introduce stability without requiring massive rewrites. Attendees will leave with concrete, proven techniques for understanding, modifying, and improving legacy code bases with confidence.