Dev Finds Software Engineering Lessons Outweigh ML Insights in Ad Click Project

A developer building an ad click prediction project discovered that software engineering challenges — not machine learning — dominated the learning curve. Before reaching model training, they encountered issues with project architecture, module imports, packaging, and test configuration. Moving data-loading logic from a Jupyter notebook into a reusable module and adding explicit error handling revealed gaps in the project's structure. When automated tests failed due to Python's module resolution, the developer found that installing the package in editable mode was impractical in their Anaconda environment, prompting a simpler fix via a pytest configuration file. The experience underscored that practical engineering often means choosing a workable solution over a theoretically ideal one.