AI-native engineering
I design, build, test, and deliver software with AI woven into my day-to-day — from research and planning to code review and delivery updates — while ownership, quality, and visibility stay in my hands.
My AI-assisted workflow
Why it matters
AI speeds up research, boilerplate, and testing — the engineering judgment behind every decision stays mine.
Planning, QA, documentation, and delivery visibility are part of how I work — not afterthoughts.
Regular, honest updates on progress, blockers, and what's next — no surprises at delivery.
Every AI-assisted output goes through my own review before it ships. I stay accountable for the outcome.
AI-assisted planning helps me scope work accurately before I commit to a timeline.
AI-assisted testing and review catch issues earlier, without replacing careful engineering judgment.
Process
Six stages, one workflow — AI supports each of them, judgment drives all of them. Click a stage to see it in practice.
In practice
I'll ask Claude or ChatGPT for a fast first pass on a new API, library, or approach — but I don't take it at face value. Anything I'm going to build on gets checked against the real documentation before I write a line of code.
Tools: Claude, ChatGPT, Perplexity
In practice
I'll paste in a feature description and ask for a task breakdown, then adjust it against what I actually know about the codebase. It's a first draft I edit, not a plan I follow blindly.
Tools: Claude, ChatGPT
In practice
GitHub Copilot handles inline completions while I'm typing. For bigger chunks I describe what I need and review the output line by line before it goes in. I don't ship code I don't understand.
Tools: GitHub Copilot, Claude, Cursor
In practice
I've got AI checks running in my CI pipeline and on every pull request. None of that replaces my own review — I triage every flag myself, since not everything it raises is worth acting on.
Tools: Claude, GitHub Copilot, CI/CD checks
In practice
PR descriptions, changelogs, and README updates start as an AI-generated first pass based on my diff and commit messages. I edit for accuracy, but documentation actually happens instead of being skipped.
Tools: Claude, GitHub Copilot
In practice
When it's time to report progress, I'll turn completed tickets into a clear, non-technical summary with AI, then edit the tone and add context only I have.
Tools: ChatGPT, Claude
Stack
The mix changes per project. Tap a tool to see what it's actually for.
GitHub Copilot
Inline completions while I'm actively coding, plus automated PR checks that flag vulnerabilities, code smells, and improvement suggestions before a human looks at the diff.
Tool choice adapts to each project's stack and constraints.
FAQ
No — every AI-assisted change goes through the same review process as anything else I write. AI speeds up the first draft, not the standard.
That's fine — I scope around it. AI is how I work faster, not a requirement for the project.
No. I use it to explore options and stress-test ideas, but architecture and technical direction stay mine.
The same way I keep any code reliable: testing, review, and holding it to the same bar regardless of how it was written.
Mostly GitHub Copilot and Claude for coding, and ChatGPT or Perplexity for research — the mix shifts by project.
Let's talk about your project, your stack, and where AI can actually help.
Schedule a call