I'm a civil engineer who works mainly with physical engineering, but I also handle automation tasks involving quantities and structural work. I know some Python and can generally follow code, but I'm worried that I have an illusion of competence because I learned programming topic by topic years ago without building many complete projects.
When I start a new coding project, I often go blank and don't know what the first step or even the first line should be. I sometimes use AI for research, explanations, and code generation, but I want to become capable of debugging, improving, and writing the code myself rather than blindly accepting AI-generated code or "AI slop."
What's the best way for a working professional to rebuild their fundamentals, get better at starting projects, and use AI as a helpful tool without becoming dependent on it?
3 Answers
Build a few small, complete projects from scratch instead of studying isolated topics indefinitely. Since you already work in engineering, useful examples could be a spreadsheet-processing tool, a calculator for a routine engineering task, or a program that extracts and organizes information from drawings or reports.
For the first attempt, try designing and writing the solution yourself. Afterward, use AI for research, explanations, code review, and suggestions for edge cases. That way you practice the fundamentals while still benefiting from AI. The important part is that you can explain what each section does, test it with known inputs, and fix it when something fails.
You don’t need to reject AI completely. Use it according to your goal: if you’re learning fundamentals, write the first version yourself and ask AI for hints rather than a complete solution. If you’re trying to deliver a practical tool quickly, generated code can be useful, but you still need tests, validation, and enough understanding to review it.
A good exercise is to take any AI-generated function and rewrite it without looking at the answer, then compare the two versions. Also practice deliberately breaking the code and diagnosing the problem. That builds the debugging skill you’ll need in real projects.
When you feel stuck, don’t begin by trying to write the whole program. First describe the task in plain language, list the inputs and expected outputs, and split the work into the smallest steps you can identify. Then write pseudocode or even a few comments before writing Python.
For example: load the file, validate the data, perform one calculation, save the result, and test each step separately. This gives you a starting point and makes debugging much easier. You can also ask AI to critique your plan or explain an error, but avoid asking it to generate the entire project until you understand the design.
That makes sense. I learned Python years ago but mostly studied individual topics rather than actually finishing projects, so starting with small tools should help me get past that blank feeling.