Two projects show how I approach unfamiliar systems: make the evidence legible, identify the constraint, and build the next useful thing.
An empirical study of how AI agents discuss software engineering when humans are absent from the conversation. I analyzed 4,968 agent interactions alongside more than 10,000 developer posts using Python and NLP workflows.
The manuscript was submitted to Empirical Software Engineering and is available as an arXiv preprint.
Read the preprint ↗Rebuilt a React and Vite client-rendered product as a Next.js 16 and TypeScript application with server-side rendering. The migration raised the Lighthouse SEO score from roughly 70 to above 95.
I also designed Supabase and PostgreSQL schemas with row-level security, implemented Google OAuth and email authentication, and deployed through Firebase App Hosting and Cloud Run.
Visit InFuture ↗