I Build in the AI Era
AI tools aren't a shortcut for me — they're part of how I think and work. Here's how that looks in practice.
Cursor-First Development
I use Cursor as my primary IDE. Composer mode, codebase context, and custom .cursorrules let me scaffold features, refactor confidently, and move faster without sacrificing quality.
My workflow starts in Cursor. I maintain project-specific .cursorrules files that give the AI context about architecture patterns, naming conventions, and project structure. This means every suggestion is aligned with how the codebase actually works — not generic boilerplate.
Claude for Architecture & Review
I use Claude not just for code generation but for talking through system design, reviewing API structures, and catching blind spots. It's like having a senior engineer available 24/7.
Before building a feature, I'll often talk through the architecture with Claude — data models, API contracts, edge cases. After implementation, I use it for code review. This two-pass approach catches issues early and produces cleaner code.
Building AI Features
I'm actively learning to integrate LLMs into full-stack applications — streaming API responses, building RAG pipelines, and connecting Spring AI to Java backends.
I'm exploring the full spectrum of AI integration: from simple prompt-based features to retrieval-augmented generation (RAG) with vector databases. Currently learning Spring AI for Java backends and LangChain.js for Node.js applications.
QA Instincts × AI Code
My QA background means I question AI-generated code the same way I'd question any untested output. I validate, test, and verify — which makes my AI-assisted code more reliable than average.
AI-generated code is only as good as the review process around it. I write tests for AI-generated functions, verify edge cases it might miss, and apply the same acceptance criteria mindset I developed as a QA engineer. The result: AI-assisted code that's actually production-ready.
“I use AI as a force multiplier — not a replacement for thinking.”
Cursor for coding. Claude for architecture decisions and code review. ChatGPT for research. These are part of how I work every day — not novelties, but core tools in my engineering workflow.