ls ./projects
A Windows XP-inspired AI desktop companion that brings chat, on-demand screen context, and local voice input to a pixel-art cat on your desktop.
I built it with Electron, React, TypeScript, and the Pi SDK to help with what you are working on. It can read accessible window text and screenshots when asked, research public web pages, and help with everyday clipboard and file tasks. Conversations save locally, and tool activity stays visible in the chat.
View repositoryHow should an AI agent decide what to remember before it knows what it will need later? My current research studies repeated context compaction as new requirements emerge and memory becomes more limited.
This is my university research project exploring how agents can preserve useful context across long-running tasks. I am developing mathematical models and experiments to understand what to keep, what to forget, and when to recover information. The work is still in progress.
View repository
A Chrome and Edge extension for attaching written or spoken feedback to specific page elements, then handing that context to a coding or research agent.
I built it to turn design and research observations into actionable notes with selected text, screenshots, and target context. Notes stay in the browser, can be grouped into review sessions, and export as Markdown or a ZIP bundle for an agent handoff.
View repository
A Python-supported macro research report on how industrial policy, strategic controls, and national bottlenecks intersect with artificial intelligence.
I combined quantitative research with written analysis to frame the central thesis, develop forecastable claims, and connect AI progress with the policy and infrastructure surrounding it.
Read the report