A few things I've built recently.

A Node.js CLI agent inspired by the MemGPT paper that experiments with working memory, recall storage, eviction, and summarized context for long-running conversations.

A small Node.js experiment that makes tool calling visible: the model returns fixed command tokens, the CLI parses them, asks for confirmation, and executes CRUD-style directory actions locally.

A vision-first AI reading assistant that acts like a "Cursor for textbooks"—understands pages in real time and provides contextual explanations with minimal friction.

Pipeline that converts flat images into layered PSDs using segmentation masks from Meta’s Segment Anything Model and scripted GIMP integration.

React-based web app for analyzing and discussing YouTube videos through an interactive, chat-like interface with optional spoiler control.

Telegram bot that logs expenses from natural-language text or voice notes, parses them with an LLM, and writes structured rows to Google Sheets.