If the codebase lacks tests or has no tests at all, start by understanding the code and write automated tests. These tests will boost our confidence when identifying performance bottlenecks and refactoring inefficient code. If automated tests exists, start by reviewing them. As they act as living documentation and helping us understand the expected behavior and outputs while gaining insights into the codebase. Once we’re familiar with it, we can safely tweak and optimize the inefficient parts of the code

Start by adding integration tests to ensure existing functionality remains intact. These tests should cover all product requirements at a high level, without delving into implementation details. Benchmark your code to identify bottlenecks, and focus on optimizing the most time-consuming sections while keeping changes localized to avoid new bugs. After each optimization, add tests to confirm the new code works as expected. Continue this process, ensuring previous tests remain intact. Finally, don’t hesitate to make project-wide changes, as your incremental optimizations and added tests will help safeguard existing functionality.

Code performance can be improved without breaking the original functionality by: - ensuring algorithms and data structures are optimised, sometimes switching can significantly improve performance - multithreading can make use of multi-core processors so tasks can be executed concurrently - by optimising database queries, using indexes, avoiding unnecessary joins and retrieving only data needed - optimising memory management by ensuring objects are released when not needed and reducing memory leaks

As a software engineer, a good way to make sure you are not unintentionally breaking a good working code is by first understanding the codebase, writing the cases by automated test whether by unit test with mocking or full end to end integration test.