What’s the best way to get better at reading unfamiliar codebases?

Treat the code like a puzzle and take baby steps. Begin with one function, one data structure, or one execution path. It’s normal not to understand every surrounding piece immediately; make a note of what you don’t know and keep following the parts that connect to your current question. The more different codebases you read, the more recurring patterns you’ll recognize, and the faster you’ll form a mental model of the next project. Simpler, well-structured programs can be especially good practice before tackling a huge production system.

A lot of it comes down to practice, but the kind of practice matters. Don’t try to understand a large project all at once. Start with a concrete task, such as tracking down a bug or changing one small behavior, then follow the relevant path through the code. Find the entry point, trace the calls, inspect the data being passed around, and gradually expand your view. Drawing the flow on paper or explaining it aloud can make complicated logic much easier to hold in your head.

Learn the common shapes of the language and domain. For example, in a typical Java web application you might look for controllers, services, and repositories. In a frontend project, you may trace pages, components, state management, and event handlers. These conventions give you a useful hypothesis about where things are, even when each project organizes them slightly differently. Good editor skills also help a lot: jump to definitions, search for references, search globally for an endpoint or piece of text, and use debugger tools or call stacks to follow execution.

Teaching and modifying code are both excellent ways to understand it. Explain a small section to someone else, rewrite or reformat a limited part while testing that behavior stays the same, or use a library you already depend on and trace how one feature works. Set breakpoints, step through interesting paths, inspect state, and deliberately make a small change to see what breaks. That turns passive reading into an investigation and helps reveal the original design and assumptions.