Understanding unfamiliar code
Use focused questions to explore structure and possible dependencies, then compare the explanation with the actual source.
Technology in practice
AI-Assisted Engineering.
ChatGPT, engineering judgment & practical validation
AI can help explore a problem, draft an implementation, or make technical material easier to work through. The value comes from combining those tools with an engineer who understands the system, reviews the changes, and checks the result.
- Frame the taskContext · Constraints · Expected behavior↓
- Explore and implementAlternatives · Draft · Review↓
- Validate the resultChecks · Behavior · Human ownership
ChatGPT
Git
TypeScript
The business context
Modern tools with an engineer responsible for the result.
I use AI-assisted tools within my engineering workflow. The work still starts with understanding the problem and ends with reviewing the implementation and checking what it does. Generated code is a proposal that needs the same care as any other change.
Implementation support
Explore alternatives and draft bounded changes with explicit requirements, existing conventions, and maintenance needs.
Review and explanation
Identify scenarios worth checking and make technical decisions easier to discuss with developers and business teams.
The layers of the stack
The tools and how I use them.
ChatGPTEngineering assistanceChatGPT
ChatGPT can help with explanations, alternatives, and drafts. I use it as part of a workflow that includes my own technical judgment. A convincing explanation or code sample still needs comparison with the project, documentation, and observed behavior.
RequirementsDefine a useful boundary
Task context
Clear instructions describe the problem, expected behavior, constraints, and examples. Smaller, concrete tasks make it easier to assess whether a suggestion fits. I avoid treating missing information as permission to invent a system’s behavior.
GitUnderstand the changeCode review
Reviewing the diff helps identify unintended scope, assumptions, dependencies, and removed behavior. The reviewer needs to understand why the change belongs in that part of the system and how it interacts with existing paths.
TestingCheck the result
Validation
Builds, targeted tests, and checks of affected workflows provide evidence about a change. The right check depends on the task. Generated test code also needs review; passing a check that repeats an incorrect assumption does not establish the intended behavior.
My approach
Keep the reasoning connected to evidence.
I give AI tools a focused task and the relevant constraints, then review the suggestions against the code and expected behavior. I remain responsible for the implementation and its validation.
Experience in context
Related project work.
My engineering practice
My approach combines AI assistance with hands-on implementation, code review, and testing. The About page describes the working qualities behind that process: communication, reliability, practical judgment, problem-solving, and teamwork.
Read about my approachAscendo Solutions website
This website combines an existing licensed design with custom content, Astro components, and reviewed changes.
Explore the website’s technical foundationOfficial documentation
Related topics.
Let’s talk about your project.
Tell me what you’re building or what needs fixing.
Start a conversation